One possible reaction to Apple's gigantic tax-optimized share repurchase program is to think that spending a lot of time fiddling with how to optimize your share repurchase program mightmean you're out of better ideas. You can ponder whether this Intel share repurchase trade described in a Lehman Brothers bankruptcy lawsuit filed yesterday supplies any evidence on that question. Intel decided to buy back $1bn of its stock in August and September of 2008, and rather than just buy it in the market it entered into a pretty fiddly forward contract with Lehman like so:1
- Intel gives Lehman $1bn on August 29.
- Lehman hands the $1bn back to Intel for safekeeping - it's Lehman's money, but Intel keeps it as collateral.
- On September 29, Lehman gives Intel some shares, based on the average price of Intel stock from August 29 to September 26.2
- The dollar amount of shares Intel buys is $1bn, if the average price is $21 or below, or $250mm, if the average price is $25 or above, or some amount linearly in between if the average price is between $21 and $25:
- If the dollar amount Intel buys is less than $1 billion, Lehman gives back the extra money.
- So in other words as the stock price goes up Intel buys fewer shares, and vice versa, which is kind of wrong-way for them3 but right-way for Lehman.
- In exchange for that risk Lehman agrees to give them a discount of 10.6 cents per share.4
- The number of shares Intel buys is equal to the dollar amount divided by the average price minus 10.6 cents:
- When they entered the contract, on August 1, Intel was at $22.35; on August 29, when the averaging started, it was at $22.87.
But then Intel's stock price dropped rather sharply over the next two months, because all stock prices dropped rather sharply, because - well, among other things, because Lehman went bankrupt on September 15. So the average price ended up being $19.8872 and the number of shares was 50,552,943.5 But on September 29, when it was due to get the shares and give back Lehman's collateral, Intel's stock price was just $17.27, making those shares worth only about $873mm. That $873mm was less than the $1bn of collateral that Intel was holding for Lehman.
Intel, sensibly enough, decided it would rather have $1 billion of cash than $873mm worth of its own stock. So it decided to DK the trade and keep the collateral instead. This was not exactly sporting - it probably wasn't exactly legal - but Lehman ... I guess had bigger problems? In any case they seem to have been okay with this for five years, and only got around to suing about it yesterday.
Lehman's basic argument is that it didn't owe Intel $1bn, it owed 50.55mm shares, and so Intel had no right to seize its collateral instead of taking the shares. This is surely right. The question of damages is a harder one; basically Intel really ought to give Lehman back the difference between (1) $1bn and (2) the value of 50.55 Intel shares. But, when? On September 29, 2008, those shares were worth $873mm, so Intel should owe Lehman's estate some $127mm. Lehman's lawyers have a clever argument that Intel was required to cover its missing shares in the market, and that it couldn't do so until November 2008 because it was in an earnings blackout, and if it had bought in the market over the course of November 2008 then it would have paid only $688mm for the shares and so I guess Intel owes Lehman $322mm.6 On the other hand, I suppose if I were Intel I'd say "okay, fine, we'll give you back your $1bn with interest, you just give us 50.55 million Intel shares." As of today those shares would seem to be worth about $1.2bn, making it basically a wash.7
We talksometimesabout how bad companies are at share repurchase and this Intel trade is no exception. Intel didn't just decide to buy back a billion dollars of its stock six weeks before the market fell off a cliff in 2008:8 it decided to do so with a structured trade that amplified its risk, buying more stock if the market crashed and less if it went up. In exchange for this, Intel got a ten cent discount on its share price. It saved about $5 million - ten cents times 50mm shares - by entering a trade that ended up losing it $127 million.9
Or would have if it'd actually completed the trade. While Intel's contract was sort of wrong-way - the lower the market goes, the more shares it buys back at an average price incorporating earlier higher prices - it had one, probably fortuitous but pretty important, right-way element. If the market went up, Intel would buy shares at the average price over the month of September, which would probably turn out to be a good price. If the market crashed, Intel would buy more shares at that average price, which would turn out to be too high. But if the market crashed and Lehman went bankrupt, then Intel would just take its billion dollars back and dare Lehman to sue it - saving $1bn that it would otherwise have spent buying back stock at pre-crash prices. So far that bet has worked out well for Intel.
Lehman v. Intel complaint [via Bloomberg]
- Exhibit 3 - forward confirm
- Exhibit 4 - CSA
- Exhibit 5 - termination letter
- Exhibit 6 - loss calculation
1.Most of this is from the confirm helpfully filed with the lawsuit as an exhibit. Intel does not seem to have publicly disclosed it, since a $1bn share repurchase for Intel is kinda small.
Here in this footnote I'll say what's obvious to the, like, twenty people who do these trades, one of whom is sort of Harvey Schwartz: this is an "accelerated share repurchase" without the acceleration. Companies often do trades similar to this to accelerate the accounting benefit of buying back stock, as well as to buy shares at a discount to VWAP. Intel's approach - which is intraquarter and collateralized, has no accounting benefit, and is done just for the discount - is somewhat less common. Of course if you did the normal sort of ASR with Lehman in August 2008, in which you prepaid the purchase price and Lehman didn't post collateral, you got screwed.
2.Actually Lehman can end the averaging period as early as September 22, giving it a bit more optionality.
3.Why is this wrong-way? Simplistically, if you're buying stock and your stock craters, you might decide to buy back more stock at the new, lower price - but not at the one-month average price that reflects all of those days when your stock was higher. This contract in effect says "if your stock crashes, you'll buy more shares at the new lower price - but you'll also go back in time and buy more shares at the old, higher price." If you had a time machine you would not, on your own, choose to do that.
4.The discount is actually on a sliding scale but it turned out to be 10.6 cents per this letter from Intel also filed with the lawsuit.
5.Again from that letter. I get like 100 more shares than they do but I guess that's rounding.
6.From the complaint:
Furthermore, Intel is subject to securities regulations in connection with purchases of its own stock and in accordance with Section 10(b) of the Securities Exchange Act of 1934 and Rule 10b-5 promulgated thereunder, cannot trade in its securities while in the possession of material non-public information. Accordingly, Intel could not have purchased its shares from the open market on September 29, 2008 and for a certain “blackout period” thereafter while it was in the possession of material non-public information. On information and belief, this “blackout period” ended on November 4, 2008. Upon information and belief, Intel could have acquired 50,552,943 shares of its own common stock in a related trade over the period from November 5, 2008 through December 1, 2008, for approximately $688 million, far less than the $1 billion it seized.
That seems wrong but I still admire its cleverness.
7.Ooh I'm sure that's not right either - not really how Loss works under ISDA - but, still, it's what I'd say.
8.In all, Intel "repurchased 324 million shares of common stock at a cost of $7.1 billion" in 2008, for an average purchase price of around $22. Intel ended 2008 at $14.66, and didn't see $22 again until March 2010. In 2009, when INTC was cheap, it bought back about 88 million shares, and all in the third quarter when it was less cheap.
9.On the simplest and probably most defensible math, just the amount Intel paid ($1bn) minus the value of its shares on delivery date ($873mm).