We say that a grantor retained annuity trust that outperforms the Internal Revenue Code Section 7520 rate will succeed; in other words, it will transfer property in a taxefficient manner. Conversely, we say that a GRAT that fails to outperform the rate will fail. But like much that is commonly said, this accepted wisdom is at best an oversimplification. However, it is not an oversimplification to note that GRATs are extremely sensitive to investment.^{1,2}
So, back in the good old days of the mid to late 1990s, when the Dow was experiencing 20 percent to 30 percent annual growth, the S&P 500 was doing as well and the NASDAQ composite was through the roof, many of the GRATs that advisors told clients to create were flying high, and those advisors looked like geniuses. In the last few years, however, as the markets have wreaked havoc on many equity portfolios, some of these clients may not have been thinking of their advisors with the same degree of reverence.
Now that IRC Section 7520 rates are at or near alltime lows, many advisors are recommending that clients create GRATs, many of which have relatively long terms so as to lock in the low Section 7520 rate for years to come. But even if the recession is over, as many economists promise, the vagaries of future capital markets will have something to say about whether these new GRATs will succeed, and may influence what the clients who create them think of their advisors a couple years down the road.
So the key questions become: How can advisors lock in the performance of a good GRAT before it goes bad? And, if a GRAT does go bad, how can they rescue their clients?
The answers: The GRAT Bailout and the ReGRAT Hedge lock in the success of a good GRAT, improve the performance of a GRAT that later goes bad, and provide results equal to a GRAT that stays good. Moreover, the GRAT Bailout eliminates the mortality risk with respect to the realized benefit in the original GRAT.
CASE STUDY
To illustrate, let's see how this hypothetical plays out:
Say a client transferred $10 million to a 10year, fixedterm GRAT on Jan. 1, 1996,^{3} when the Section 7520 rate was 6.8 percent. The GRAT is economically zeroedout (that is to say, the present value of the annuity payments is equal to the value of the property transferred to the GRAT). The annuity payments increase by 20 percent each year.^{4} Also assume that the investment performance of the GRAT has tracked exactly the performance of the Dow Jones Industrial Average during the relevant period; for example, assume that the GRAT's growth rate for 1996 was equal to the difference between the close of the DJIA on Jan. 1, 1997 ($6,448.27) and Jan. 1, 1996 ($5,117.12), or about 26 percent.
As a result, the investment performance of the hypothetical GRAT was very strong in its first four years (1996 through 1999), but poor in years five, six and seven (2000 through 2002). A GRAT with these characteristics will have about $9.25 million on hand as of Jan. 1, 2003, immediately after the satisfaction of the seventh annuity payment. (See “GRAT's Performance, Years 17,” page 30.) The GRAT still must satisfy three annuity payments, each of which will be 20 percent larger than the preceding one.
To illustrate a few points at once, assume that the hypothetical GRAT continues to do poorly, realizing losses of 3 percent, 2 percent and 1 percent in years eight, nine and 10, respectively. The average annual rate of return of the GRAT over the 10year period is 5.39 percent, or about 140 basis points less than the Section 7520 rate (6.8 percent) when the GRAT was created. The GRAT remainder is equal to almost $1 million at the termination of the GRAT. (See “GRAT Transfers $1 Million,” page 30.)
Thus, the hypothetical GRAT succeeds in transferring almost $1 million to the remainderman — about 10 percent of the initial principal. That's not bad, considering that the GRAT had six consecutive down years. But the grantor may not be thrilled, because in 1999 or 2000 he probably anticipated greater success.
It's interesting to consider why this GRAT succeeded. If we believed the commonly accepted wisdom, we might expect the GRAT to fail, given that its average annual rate of return is significantly less than the relevant Section 7520 rate. But in fact, the GRAT's extraordinary early returns created a cushion that enabled it to satisfy subsequent annuity payments while weathering investment losses: Taking into account the pattern of the annuity payments, the illustrated GRAT actually achieves an internal rate of return of 7.62 percent, well in excess of 6.8 percent, the Section 7520 rate.^{5}
Now what if you had run into our hypothetical client at a cocktail party near the end of 2002, after three years of poor performance, and he said: “That GRAT was awesome at first, and even now it's worth a lot more than when I funded it. But if the market continues like this, I'm going to wind up giving it all back. It's still got a long time to run. And who knows when I'm going to die. Is there any way to cash out of this thing?” The answer is: “Yes.”
The grantor can buy the GRAT remainder from the remainderman at the end of year seven (on Jan. 1, 2003) for its thenactuarial value. Such a transaction should immediately transfer the thenpresent value of the GRAT benefit to the remainderman free of gift tax.^{6}
Under Chapter 14, the present value of the remainder should be equal to the total value of the GRAT property less the present value of the remaining annuity payments, using the current Section 7520 rate as the discount rate.^{7}
In our example, the total value of the GRAT at the end of year seven is $9.25 million. (See “GRAT's Performance, Years 17,” page 30.) The present value of the remaining annuity payments would be equal to about $7.15 million, using a discount rate of 4.2 percent (the Section 7520 rate for January 2003).^{8} (See “Present Value of Remaining Annuity Payments,” this page.)
Thus, at the end of year seven, the right to receive the remainder of the GRAT when it terminates at the end of year 10 has a present value of about $2.1 million ($9.25 million — the value of the GRAT at the end of year seven — less $7.15 million — the present value of the remaining annuity payments — equals $2.1 million).
Accordingly, if the grantor wanted to purchase the remainder, he would buy it from the remainderman^{9} for $2.1 million, thus locking in the thenpresent value of the benefit of the GRAT thus far.^{10}
If the remainderman invested that amount for three years, and the investment performance of the property in the remainderman's hands was the same as the GRAT illustrated in the example (losses of three percent, two percent and one percent in years one, two and three respectively), at the end of the third year (that is, at the end of the total 10year period), the remainderman would have approximately $1.98 million. (See “Remainderman's Side Fund,” this page.)
By contrast, if the grantor had not entered into the transaction, the remainderman would have received slightly less than $1 million at the termination of the GRAT. (See “GRAT Transfers $1 Million,” page 30.) Thus, as a result of the sale of the remainder to the grantor, the remainderman receives almost $2 million, about $1 million more than if the grantor had not purchased the remainder. The remainder purchase has the effect of transferring approximately 20 percent of the initial GRAT principal to the remainderman — a rather spectacular result for a GRAT that spent six of its 10 years losing money!
In addition, the remainder purchase protects the remainderman from the mortality risk inherent in a GRAT. If the grantor dies in year nine, for example, the amount the remainderman receives is not included in the grantor's estate and is not otherwise affected by the grantor's death.
In this example, the grantor called it right. The GRAT performed poorly in years eight, nine and 10. But what if his forecasting was wrong, and investment performance over the remaining term of the GRAT had been strong?
THE REGRAT HEDGE
Let's say that the investment performance of the GRAT would have experienced a turnaround, and would have realized growth of 10 percent, 12 percent and 14 percent in years eight, nine and 10, respectively. If so, there would be about $4.2 million in the GRAT when it terminated. (See “Better Than Expected,” page 33.)
But as we have postulated, the grantor purchased the remainder at the end of year seven for $2.1 million. If the remainderman were to match the performance of the GRAT (growth of 10 percent, 12 percent and 14 percent in years eight, nine and 10, respectively), at the end of year 10, the remainderman would have only $2.9 million, about $1.3 million less than if the GRAT had continued. (See “Remainderman's Side Fund,” page 32.)
Obviously, that is not a good result. So maybe the remainder purchase is beneficial only if the GRAT does not outperform the grantor's predictions. Or maybe there's a way to structure it so that it makes sense no matter which way the wind blows.
Taking a step back, as a result of the remainder purchase, the grantor owns every beneficial interest in the trust: He already was the beneficial owner of the annuity interest and, with the purchase of the remainder interest, he becomes the beneficial owner of the remainder interest. So from the perspective of beneficial interest, the grantor owns the whole ball of wax. Under those circumstances, it should be possible for the grantor to purchase all or a portion of the property of the GRAT, in exchange for his promissory note.^{11} He then could contribute to a new GRAT a portion of the GRAT property with a value equal to the present value of his annuity interest in the original GRAT, giving the new GRAT a term equal to the unexpired term of the original GRAT.^{12}
This new contribution, in effect a partial, accelerated ReGRAT,^{13} would serve as a hedge if the performance of the original GRAT exceeds his expectations: If the grantor guesses wrong about the investment performance of the GRAT, the combination of the purchase price the remainderman receives with respect to the original GRAT, together with the remainder of the new GRAT, will put the remainderman in at least as good a position as if the grantor had not purchased the remainder. But if he guesses right and the GRAT does poorly, the remainderman will be in a better position. In other words, by virtue of the remainder purchase and the contribution to the new GRAT, the grantor puts the remainderman in a potential win, but ultimately nolose, position.
To illustrate, let's say that, after purchasing both the remainder and the property of the original GRAT, the grantor takes the difference between the value of the GRAT at the time when he buys the remainder and the purchase price for the remainder, then contributes that difference to a new GRAT. (Of course, the difference is equal to the present value of the remaining annuity payments due from the original GRAT.) On the facts of our hypothetical, the total value of the GRAT at the end of the seventh year amounts to $9.25 million, the present value of the remaining annuity payments is equal to $7.15 million (at a discount rate of 4.2 percent, the Section 7520 rate in January 2003), and the purchase price for the remainder is $2.1 million. Accordingly, the grantor would purchase the remainder for $2.1 million. He also would purchase $7.15 million of the property of the old GRAT in return for his note, then contribute that property to the new GRAT.
Assume the grantor takes these steps, and the new, threeyear GRAT that he creates exceeds his investment expectations, returning 10 percent, 12 percent and 14 percent in years one, two and three, respectively. The remainderman would have roughly $2.9 million from the investment of the proceeds from the remainder sale based on those returns. (See “Remainderman's Side Fund After GRAT Sale,” this page.) In addition, the new, threeyear GRAT will transfer almost $1.3 million to the remainderman. (See “Performance of New GRAT,” this page.) Thus, taking into account the proceeds of the remainder sale and the value of the new GRAT, the remainderman receives a total of about $4.2 million, which is the same amount the remainderman would have received if the grantor had not purchased the remainder. (See “Better Than Expected,” page 33.) Yet the grantor has protected them from poor investment performance.^{14}
LEGAL ISSUES
Given their effectiveness, the GRAT Bailout and the ReGRAT Hedge are unusually unencumbered by thorny legal issues. The treatment of the remainder purchase under Chapter 14 seems straightforward. Where the grantor is a parent of the remainderman, and thus a member of the remainderman's family, the treatment under Chapter 14 should be:

When the remainderman sells his interest to the grantor, the remainderman is making a transfer of an interest in trust for the benefit of a member of his family.

Section 2702(a)(1) provides that, solely for purposes of determining whether such transfer is a gift (and its value), the value of any interest in the trust retained by the transferor or any applicable family member is determined under Section 2702(a)(2).

The transferor (the remainderman) retains no interest in the trust.

The grantor, as the remainderman's parent, is an applicable family member of the remainderman.^{15} The grantor does retain an interest in the trust — the annuity.^{16} However, the annuity is a qualified annuity interest.^{17} Thus, the annuity should be valued under Section 7520.

For a case in which the remainderman is a grantor trust, and the grantor is the parent of the beneficiary or beneficiaries of such trust, the result should be the same as when no trust is involved, either because the interests of the beneficiaries in the trust are attributed to them (in effect, looking through the trust), or because the transaction is treated as a sale between the grantor and the trustee, in which case Chapter 14 should have no applicability.^{18} In the latter case, there should be no question of adequate consideration but, if this were necessary in order to avoid a gift, the use of Section 7520 to value the remainder interest should avoid any gift tax consequences.^{19}

If the grantor purchases property from the original GRAT and gives the trustee his promissory note in exchange, a distribution to the grantor of a fractional interest in his note may become necessary to satisfy in whole or in part the remaining annuity payments owed by the original GRAT. If so, the payment of an interest in the note to the grantor would appear not to violate Section 25.27023(b) of the Treasury Regulations. This regulation requires, among other things, that in order to be a qualified annuity interest, the interest must be an irrevocable right to receive a fixed amount that must be payable to the holder of the interest, and provides further that the “issuance of a note … directly or indirectly, in satisfaction of the annuity amount does not constitute payment of the annuity.”^{20} The satisfaction of an annuity payment, due from a GRAT that holds notes issued by the grantor, by the inkind payment of an interest in the notes does not constitute issuance of a note by the GRAT and thus would appear not to run afoul of the regulation.
BE PROACTIVE
So there is a way out of a bad GRAT. But perhaps even more important, the GRAT Bailout and the ReGRAT Hedge give investment advisors, attorneys and other wealth planning professionals an additional opportunity to bring value to their clients. Don't wait for clients to ask. Instead, recognize, then make the most of, the fact that the planning isn't over when the GRAT agreement is signed. The opportunity to lock in the success of a GRAT and engage in hedging strategies should encourage advisors to monitor the value of their clients' portfolios inside GRATs, and add immense value long after the GRAT itself has been implemented — by suggesting these (or other investment strategies they may have up their sleeves) at the appropriate times. Lest we forget, one thing the last three years should have taught us is that capital markets are uncertain.
The author thanks his Smith Barney colleagues Robert B. Seaberg and Gary J. Nestler, and his former Kirkland & Ellis colleague Kevin M. Chen, for their helpful comments.
Endnotes

Portions of this article appeared, in a different form, in Glenn Kurlander, “Investment Driven Estate Planning: Examining the Relationship Between Wealth Transfer Strategies and Investment Performance” in 54 Major Tax Planning, Paragraph 1400 (2002), and are used here with the permission of Matthew Bender.

A GRAT with an average annual rate of return that is less than the Section 7520 rate may succeed so long as its internal rate of return, taking into account the pattern of annuity payments over the term of the GRAT, is greater than the Section 7520 rate.

In fact, prior to the decision in Walton v. Comm'r, 115 T.C. 589 (2000), fixedterm GRATs were unusual, and it was much more common for the grantor of a GRAT to retain a contingent reversion, in the event he died prior to the expiration of a fixed term of years. (In other words, it was common for the term of a GRAT to end upon the first to occur of either: the expiration of a fixed term of years, or the grantor's death; these trusts are called “shorter of” GRATs.) This is because the infamous Example 5 under Section 25.27023(e) of the Treasury Regulations took the position that an annuity payable for a term of years nonetheless had to be valued as if it had been created for the shorter of the term or the grantor's life. Given this controversial regulation, most advisors concluded that if the grantor would be taxed as if he retained a contingent reversion no matter what, then the grantor might as well retain such an interest, so that, if he died during the term, the interest could come back into the grantor's estate and be disposed of so as to qualify for the estate tax marital deduction if the grantor's spouse survived. Notwithstanding this historical reality, the illustration uses a fixedterm GRAT for the sake of simplicity, even though it was something of a historical oddity in 1996.

As a general matter it is often advantageous to structure a GRAT with annually increasing, as opposed to level, annuity payments. See, e.g., Glenn Kurlander, “Investment Driven Estate Planning: Examining the Relationship Between Wealth Transfer Strategies and Investment Performance,” 54 Major Tax Planning, Paragraph, 1400, 1402.4.D (2002). The amount of the annuity may increase annually by up to 20 percent of the preceding year's annuity. Treas. Reg. Section 25.27023(b)(1)(ii)(B).

Interestingly, the low annuity payments in the early years of the GRAT ultimately do not account for the GRAT's success. In fact, based on the investment assumptions illustrated in the text, a zeroedout GRAT with level, as opposed to increasing, annuity payments would succeed in transferring even more property to the remainderman than the illustrated GRAT. Thus, if each year's annuity was equal to $1,410,636, at the end of the 10year term, the remainderman would receive $2,419,269, rather than $998,231. (See “Grat's Performance, Years 17,” page 30). This is because the GRAT with level annuity payments would have an internal rate of return of 9.15 percent, as opposed to 7.62 percent in the case of the GRAT with increasing payments. Based on the investment assumptions described in the text, the GRAT with increasing payments has more property in it at the end of each year than a GRAT with level payments, except for the very last year, when the final annuity payment in the increasing GRAT grows so large that it consumes almost 76 percent of the property then in the GRAT (as opposed to only 37 percent of the property then in the GRAT with level payments).
Using an extreme example to illustrate the point, assume that the GRAT regulations permitted one to create a GRAT that would make no annuity payments until the end of the final year, at which time the GRAT would be required to pay the grantor an amount equal to the future value of the initial principal, assuming it had grown at the applicable Section 7520 rate. Based on a Section 7520 rate of 6.8 percent and the investment results illustrated in the text, the GRAT would be required to pay the grantor $19,306,899 when it terminated, but would have on hand only $15,341,164, and thus would fail. The internal rate of return for such a GRAT over the 10year period would be equal to only 4.37 percent, well below the hurdle rate.

The use of the phrase “thenpresent value” may seem redundant; it intentionally reflects, however, that the remainder purchase itself is sensitive to the discount rate used to compute the present value of the various interests in the GRAT (i.e., the Section 7520 rate in effect when the remainder purchase is consummated) and that, in an absolute sense, computing the true thenrealized GRAT benefit requires the use of a discount rate equal to the Section 7520 rate in effect when the GRAT was created. Such a historical confluence will occur only by coincidence. In the case of longterm GRATs that have been in effect for some time, such a coincidence is unlikely to occur, given the persistent decline in Section 7520 rates over the last few years to what are now historic (or nearhistoric) lows.
This also means that this technique will have added bang when the Section 7520 rate in effect at the consummation of the remainder purchase is higher than the rate in effect when the GRAT was created and, conversely, will entail a diminution of the benefit when the former is lower than the latter. I call this effect “discount rate distortion,” and one must recognize that it is a cost of the GRAT Bailout in some cases, but a boon in others.

Support for this assertion will be discussed below in the section titled “Legal Issues.”

Because the illustrated GRAT is a fixedterm GRAT, the determination of the present value of the remainder is a simple, purely financial calculation. If the GRAT is structured with a contingent reversion, however, determination of the value of the remainder requires the introduction of a mortality component as well, because there is a chance that the grantor may die during the term. This calculation is significantly more complex, and changes day by day as the grantor becomes older and his chances of dying within the term change, albeit in a small way (which can still have significant consequences when the value of the GRAT is substantial).

For the sake of simplicity, and to mirror common practice, I am assuming that the remainderman of the GRAT is another trust that, for income tax purposes, is treated as a grantor trust as to our grantor. This allows for transactions between the grantor and the remainderman without income tax consequence, and also simplifies the illustration of comparative investment performance without the need to take into account income tax effects. If the remainderman of the GRAT is not a grantor trust, the sale of the remainder interest likely will present a significant income tax cost, which in many cases can exceed the transfer tax saving achieved by the sale.

It is probably most accurate to think of “locking in” as a relative term, because the purchase of the remainder reflects a kind of arbitrage, which is subject to what I call “discount rate distortion.” In other words, to achieve a lock in of the thenalreadyrealized GRAT benefit, the purchase would have to be consummated at a time when the Section 7520 rate is equal to the rate in effect when the GRAT was created. That is because the use of a higher or lower discount rate introduces a “distortion” into the computation of the present value of the remaining annuity payments that can produce either an added benefit to the transaction or represent an additional cost.
To illustrate, the Section 7520 rate in effect when the GRAT in the example was created was equal to 6.8 percent. If that same rate were in effect on Jan. 1, 2003 when the remainder purchase takes place, and using the Section 7520 rate as the discount rate for purposes of computing the present value of the annuity payments, the present value of such payments would be equal to approximately $6.8 million, and thus the purchase price of the remainder would be equal to approximately $2.5 million (that is, $9.3 million, the value of the GRAT at the end of year seven, less $6.8 million the present value of the remaining annuity payments, computed at 6.8 percent, equals $2.5 million). Put differently, in presentvalue terms, the GRAT has achieved a benefit of $2.5 million. (As one would expect, the value of the remainder of the illustrated GRAT at the end of year 7 is $2.5 million more than the value of a GRAT measured at the same time that had earned exactly 6.8 percent per year, in each year of its existence.)
The use of the lower discount rate that actually applied in January 2003, 4.2 percent, has the deleterious effect, however, of increasing the present value of the remaining annuity payments when compared to the computation of such payments using a discount rate of 6.8 percent, and thus decreasing the value of the remainder, the amount that the grantor pays to the remainderman. As the example shows, using 4.2 percent results in a present value for such payments of approximately $7.2 million and a purchase price for the remainder of approximately $2.1 million, or about $400,000 less than if the Section 7520 rate were the same at both the time of creation and the time of consummation of the remainder purchase.
Thus, because of “discount rate distortion,” the grantor's purchase of the remainder transfers something less than the benefit accomplished thus far by the GRAT.
By contrast, if the 7520 rate is higher at the time the remainder purchase is consummated than it was at the time the GRAT was created, the technique has the effect of transferring gifttaxfree more than the thenrealized GRAT benefit, again because of discount rate distortion.
For example, if the Section 7520 rate on Jan. 1, 2003, the date on which the remainder purchase is consummated, was equal to 8 percent, the present value of the grantor's retained annuity would be $6.6 million instead of $7.1 million, and the purchase price for the remainder would be $2.6 million instead of $2.1 million. The roughly $500,000 difference is attributable solely to the difference in 7520 rates and is transferred totally taxfree.
Thus, “discount rate distortion” can represent either a boon or a cost depending upon whether the 7520 rate is higher or lower at the time of the purchase than it was at the time of the GRAT creation: As shown above, in the former case the technique has the effect of transferring gifttax free more than the alreadyrealized GRAT benefit; in the latter case, it transfers less.

The trustee should have no cause to object to the grantor's purchase of the GRAT property, because, as a result of the remainder purchase, the grantor becomes the only person beneficially interested in the GRAT.

In this way, the grantor can effect the GRAT Bailout and the ReGRAT Hedge without contributing any additional property to the arrangement. If, however, he wished to transfer additional property to the remainderman, he could transfer to the new GRAT the entire value of the original GRAT ($9.3 million on our facts), instead of just the present value of the annuity payments ($7.1 million). Of course, he could transfer any greater or lesser amount as well.

It is accelerated in the sense that the grantor contributes to a new GRAT the present value of annuity payments due in the future; he does not wait until the payments actually are made to ReGRAT them.

It should not be surprising that the results of the original GRAT, on the one hand, and the sum of the investment of the proceeds of the remainder sale, and the results of the new GRAT, on the other, are the same. That is because the GRAT Bailout and ReGRAT Hedge, taken together, represent nothing more than the division of the original GRAT into two pieces, both of which, for the sake of illustration, are shown to grow at the same rate as one another (and at the same rate as the original GRAT). Put numerically, the amount in the original GRAT at the beginning of the eighth year, $9,254,596 is equal to the sum of the amount of the new GRAT ($7,152,952), and the amount the remainderman invests outside of the GRAT in such year ($2,101,644).
From this, we may conclude that the ReGRAT Hedge eliminates the discount rate distortion to which the GRAT Bailout is subject, discussed above in note 9, because when the two techniques are combined, the same amount will continue to be invested for the benefit of the remainderman. As noted, when the Section 7520 rate in effect at the time of the remainder purchase is higher than the rate in effect when the GRAT was created, the grantor will pay more for the remainder interest, and thus the remainderman will have more to invest. When the ReGRAT Hedge is introduced, however, the net amount available to fund it is correspondingly less than it would have been if the Section 7520 rate then in effect was lower than the rate that applied when the GRAT was created. Thus, there is an inverse relationship between these amounts, such that the sum of the value of the new GRAT and the remainder proceeds is a constant.

Section 2701(e)(2)(B).

The remainder interest that the grantor acquires is not treated as a retained interest because the grantor did not hold it before and after the transfer. Treas. Reg. Section 25.27022(a)(3).

Section 2702(b)(1).

Treas. Reg. 25.25128: “A sale, exchange or other transfer of property made in the ordinary course of business (a transaction which is bona fide, at arm's length, and free from any donative intent), will be considered as made for an adequate and full consideration in money or money's worth.”

Treas. Reg. Section 25.25111(g)(1): (“The gift tax is not applicable to a transfer for a full and adequate consideration in money or money's worth.”)

Treas. Reg. Section 25.27023(b)(1)(i).
GRAT'S PERFORMANCE, YEARS 17
A 10year, zeroedout GRAT funded with $10 million on Jan. 1, 1996 (the investment performance of which exactly tracks changes in the DJIA, year over year) will have about $9.25 million on Jan. 1, 2003
CALENDAR YEAR  GRAT YEAR  BEGIN YEAR  GROWTH RATE  GAIN (LOSS)  ANNUITY PAID  END YEAR 

1996  1  $10,000,000  26.01%  $2,601,366  ($598,093)  $12,003,273 
1997  2  12,003,273  22.64  2,717,712  (717,712)  14,003,273 
1998  3  14,003,273  16.10  2,254,441  (861,254)  15,396,460 
1999  4  15,396,460  25.22  3,883,211  (1,033,505)  18,246,166 
2000  5  18,246,166  6.18  (1,127,213)  (1,240,206)  15,878,748 
2001  6  15,878,748  7.10  (1,126,631)  (1,488,247)  13,263,870 
2002  7  13,263,870  16.76  (2,223,377)  (1,785,896)  9,254,596 
2003  8  9,254,596  
2004  9  
2005  10 
GRAT TRANSFERS $1 MILLION
If the GRAT experiences losses of 3 percent, 2 percent and 1 percent in years 2003, 2004 and 2005, respectively, it still will succeed in transferring almost $1 million to the remainderman
CALENDAR YEAR  GRAT YEAR  BEGIN YEAR  GROWTH RATE  GAIN (LOSS)  ANNUITY PAID  END YEAR 

1996  1  $10,000,000  26.01%  $2,601,366  ($598,093)  $12,003,273 
1997  2  12,003,273  22.64  2,717,712  (717,712)  14,003,273 
1998  3  14,003,273  16.10  2,254,441  (861,254)  15,396,460 
1999  4  15,396,460  25.22  3,883,211  (1,033,505)  18,246,166 
2000  5  18,246,166  6.18  (1,127,213)  (1,240,206)  15,878,748 
2001  6  15,878,748  7.10  (1,126,631)  (1,488,247)  13,263,870 
2002  7  13,263,870  16.76  (2,223,377)  (1,785,896)  9,254,596 
2003  8  9,254,596  3.00  (277,638)  (2,143,075)  6,833,883 
2004  9  6,833,883  2.00  (136,678)  (2,571,690)  4,125,515 
2005  10  4,125,515  1.00  (41,255)  (3,086,029)  998,231 
PRESENT VALUE OF REMAINING ANNUITY PAYMENTS
On Jan. 1, 2003, the GRAT still must fund three annuity payments; using a discount rate of 4.2 percent (the Section 7520 rate in January 2003), the sum of the present values of such payments is equal to about $7.15 million
CALENDAR YEAR  GRAT YEAR  ANNUITY PAID  PRESENT VALUE 

2003  8  $2,143,076  $2,056,695 
2004  9  2,571,690  2,368,555 
2005  10  3,086,029  2,727,702 
REMAINDERMAN'S SIDE FUND
If the remainderman invests the $2.1 million the grantor pays to buy the remainder, and experiences losses of 3 percent, 2 percent and 1 percent, the remainderman will have about $1.98 million
CALENDAR YEAR  BEGIN YEAR  GROWTH RATE  GAIN (LOSS)  END YEAR 

2003  $2,101,644  3.00%  ($63,049)  $2,038,595 
2004  2,038,595  2.00  (40,772)  1,997,823 
2005  1,997,823  1.00  (19,978)  1,977,845 
BETTER THEN EXPECTED
What if the GRAT did better than the grantor expected? If the grantor's fears were not realized, and the GRAT experienced growth of 10 percent, 12 percent and 14 percent in its final three years, the remainderman would receive more than $4.2 million
CALENDAR YEAR  GRAT YEAR  BEGIN YEAR  GROWTH RATE  GAIN (LOSS)  ANNUITY PAID  END YEAR 

1996  1  $10,000,000  26.01%  $2,601,366  ($598,093)  $12,003,273 
1997  2  12,003,273  22.64  2,717,712  (717,712)  14,003,273 
1998  3  14,003,273  16.10  2,254,441  (861,254)  15,396,460 
1999  4  15,396,460  25.22  3,883,211  (1,033,505)  18,246,166 
2000  5  18,246,166  6.18  (1,127,213)  (1,240,206)  15,878,748 
2001  6  15,878,748  7.10  (1,126,631)  (1,488,247)  13,263,870 
2002  7  13,263,870  16.76  (2,223,377)  (1,785,896)  9,254,596 
2003  8  9,254,596  10.00  925,460  (2,143,075)  8,036,980 
2004  9  8,036,980  12.00  964,438  (2,571,690)  6,429,728 
2005  10  6,429,728  14.00  900,162  (3,086,029)  4,243,861 
REMAINDERMAN'S SIDE FUND AFTER GRAT SALE
Assuming the GRAT had exceeded the grantor's expectations, if the remainderman invests the $2.1 million the grantor pays to buy the remainder, and experiences gains of 10 percent, 12 percent and 14 percent, the remainderman will have about $2.9 million in 2005, which is $1.3 million less than the remainderman would have received from the GRAT had the grantor not purchased the remainder
CALENDAR YEAR  BEGIN YEAR  GROWTH RATE  GAIN (LOSS)  END YEAR 

2003  $ 2,101,644  10.00%  $ 210,164  $ 2,311,808 
2004  2,311,808  12.00  277,417  2,589,225 
2005  2,589,225  14.00  362,492  2,951,717 
PERFORMANCE OF NEW GRAT
If the grantor transfers $7.15 million to a new, threeyear GRAT that has returns of 10 percent, 12 percent and 14 percent in years one, two and three, respectively, the remainderman will receive almost $1.3 million from the new GRAT
CALENDAR YEAR  GRAT YEAR  BEGIN YEAR  GROWTH RATE  GAIN (LOSS)  ANNUITY PAID  END YEAR 

2003  1  $ 7,152,952  10.00%  $715,295  ($2,143,078)  $5,725,169 
2004  2  5,725,169  12.00  687,020  (2,571,694)  3,840,496 
2005  3  3,840,496  14.00  537,669  (3,086,032)  1,292,133 