A hedge fund friend who owns a couple of million shares of Micron Technology (NASDAQ:MU) is fond of reminding me, "The only perfect hedge is in a Japanese garden." Alas, to the huge detriment of Micron's balance sheet, the company's CFO Ron Foster has designed a convertible bond which can be almost perfectly hedged by the convertible bond arbitrageurs who own these securities. These securities have been great for the arbs and terrible for the company. This article explains why Mr. Foster should be quickly retired to a Japanese garden, what exactly the damage is that he has wrought, and what could and should be done about it-- quickly-- to avoid further damage.
Converts "101." What are these things? I took a swag at trying to explain what a convert is here. So did SA's Retired Securities Attorney, here. Too lazy to click on the links? Basically a convertible buyer buys a bond with $1000 face value. The bond indenture states how many shares into which the bond may convert, at the holder's option. The convert buyer then sells short that number of shares. Presto! No net dollars invested! Now they just settle down for say the next 29 years (in the case of the most recent convert Micron issued) and collect $30 in interest per $1000 of face value in bonds they have purchased.
Quiz review: a) spend $1000 investing in a bond, b) receive $1000, deposited in your account, from short selling underlying shares, c) $30 is received each year on no net money invested.
Exam Question: TRUE or FALSE: $30 per year for 29 years on no net money invested is a good deal for the bondholder?
Neither RSA nor I really hammered home that this transaction isn't about a beauty contest for DRAM or NAND. The convert arbitrageur may not be able to spell those memory types, let alone care about them. It is not about whether Micron is better than SanDisk (SNDK) (who have also done converts by the way....). The arb doing this transaction doesn't even really care what Micron's earnings are within a very broad range. This is almost a pure trading decision: How little margin can a buyer put down to buy the bond and short the stock? RSA asserts none, which would be a very big sophisticated hedge fund and broker on the other side.
Background. In order to show enough liquidity to be a credible bidder for the now company-owned Elpida subsidiary, the company felt that it needed to show a significant cash balance. The bidding started in 2012 and the acquisition closed in the summer of 2013. Apparently, in order to avoid restrictive covenants and high interest coupons of regular debt, the company felt convertible bonds were the way to go. They offered lower interest rates than regular debt and fewer restrictive covenants. The company would even protect itself from an underlying stock price rise by purchasing "capped calls," preventing or minimizing dilution. Or so we were told. These "capped calls" were a mini-disaster of their own, as is discussed below.
The real mystery is why the company continued to issue these toxic, dilutive converts even after its stock price ignited in late 2012 and early 2013. This was also the period in which the tremendous cash flow capability of the company began to be apparent to even the most callous observers.
As shown below the company issued a lot of bonds at different times, with each group known as a tranche, which Merriam Webster defines as follows:
a division or portion of a pool or whole; specifically : an issue of bonds derived from a pooling of like obligations (as securitized mortgage debt) that is differentiated from other issues especially by maturity or rate of return
How many tranches of this stuff did Mr. Foster do? Here is a chart of convertible bonds outstanding at various times:
Face value in millions, including current portion:
|Rate||Tranche||2012||2013||2Q 2014||3Q 2014|
-2014 3Q pro-forma is author's estimate that 3Q redemptions happen on announced schedule.
-blank cells are for tranches not yet issued, or tranches repurchased or exchanged.
So that's it for the amount of convertibles issued: $1.981 Billion proforma for the already scheduled redemptions? No, not quite. This is the amount which ties to the balance sheet and is a good example of "GAAP is crap." It's a trap I fell into in an early draft of this article. One must dig deeper in the most recent 10Q, back on page 17, to find this chart:
So what's the difference between this slide showing $2.601 billion outstanding and the previous GAAP slide showing $1.931 outstanding? It is principally the OID (original issue discount) on the 2043G Notes and the fact that GAAP books part of these somewhat equity-oriented securities outside of the debt portion of the balance sheet. From here on, this article will use these numbers for each tranche. Still the previous table is interesting to see the evolution of the convertible notes outstanding. I applaud the progress redeeming notes for cash, as is extensively described in the footnotes to the 10-Qs and 10-Ks. I abhor the issuance of the 2043G tranche, which was offered in exchange for other convertibles, since at best this is "kicking the can down the road" and at worst is like fighting a fire with gasoline.
The table from the 10Q shown immediately above is very important for one other reason. The last column is entitled "Conversion Value in excess of Par." This is one of the few places Micron gives a hint that these converts have ballooned in market value, due to the soaring stock price and the implicit underlying conversion value as will now be described in some detail.
What is this stuff worth in the market today? The problem for Micron is that these converts grow as the stock price grows. And with converts they don't sink if the stock price sinks below the initial conversion price; they stay at par.
The prices below were provided by a friend on an institutional bond desk and reflect a closing stock price of $22.60. Stock price is critical since these bonds are so sensitive to their conversion value. Quotes on converts aren't often widely available online or via many quoting services. As will be seen below, these prices are very close to the "as converted" stock value:
|Face Value||Market Value|
|Rate||Tranche||2Q 2014||Quote||for Tranche|
|1.875%||2031B||114||239.88 x 240.63||274|
|2.375%||2032C||450||241.375 x 241.875||1,087|
|3.125%||2032D||412||239.0 x 239.5||986|
|1.625%||2033E||300||212 x 213||638|
|2.125%||2033F||300||214.25 x 215.25||644|
|3.000%||2033G||1025||106.75 x 107.75||1,099|
- mid point between bid and ask used for market value
- Your forgetful author originally forgot to ask the bond geniuses for quotes on the 2032 tranches. Until the quotes arrived I was using the implicit value for these two tranches which is: the number of as-converted shares times the $22.60/share price implicit in the other quotes. For example, each bond in the 2032C tranche converts into 103.8907 shares. Multiplying those numbers, one gets $2347.93 which would be equivalent to 2.34793 in the table above. The difference would be due to the "optionality" and accrued interest primarily. So if you don't want to bother your friends trading emerging market debt and ask them to go bother their Master of the Universe convertible traders, you could always just punt and use the intrinsic value plus some fudge. Thanks for the quotes Kenny!
This table shows the problem Micron is facing: bonds have risen in value to 106% to 241% of face value due to the rise in stock price.
What about those "capped calls?" The company issued several different "capped calls" which were supposed to protect against a price rise. Here is a description of the capped calls issued in April of 2012. These were issued in conjunction with the 2032 tranches. The floor, as described below ranged from $9.80 per share to $10.16 per share; the "cap" in this transaction was $14.26 to $16.04 per share. This transaction covered 100.7 million shares. From the link cited, here is what they were supposed to do and what they cost:
The Capped Calls are intended to reduce the potential dilution upon conversion of the Notes. If, however, the market value per share of the Common Stock, as measured under the terms of the Capped Calls, exceeds the applicable cap price of the Capped Calls, there would be dilution to the extent that the then market value per share of the Common Stock exceeds the cap price. Additionally, to the extent that the market value per share of Common Stock exceeds the conversion price of the Notes but does not exceed the strike price of the Capped Calls, Micron will not be entitled to receive any shares of Common Stock under the Capped Calls. Micron paid approximately $103 million from the net proceeds from the issuance and sale of the Notes to purchase the Capped Calls. The expiration dates of the Capped Calls range from four to six years.
Here is a summary of all the capped calls, found even further back in the most recent 10Q, on page 22, far removed from the discussion of the debt they were meant to protect:
I believe I have found all of the 8-Ks dealing with the issuance of the capped calls and presently outstanding converts: 7/26/11 for the 2031 notes, 4/13/12 for the 2032 notes, 2/12/13 for the 2033 notes, and 11/18/13 for the 2043 notes. Those 8-Ks detail $194 million for the purchase of these capped calls.
So our CFO, Mr. Foster, bought 4 to 6 years of very expensive insurance for underlying instruments that had a 20++ year lives. He bought protection only up to a maximum of $16.04 per share, and which wouldn't have been triggered in other instances as cited. What kind of insurance did we get for our $194 million insurance premium? According to the chart above we got $954 million of value, which is more than I think we realized from the Elpida insurance where I don't think we realized any value. I would liken this to receiving insurance coverage for the garage and first floor of a building whose top 3+ floors have been completely destroyed. Mr. Foster delivers the near perfect "Japanese Garden" type hedge to the convertible bond arbitrageurs while he savages our own balance sheet.
What has the company told shareholders? Here's a brand new slide which the company helpfully included in their most recent earnings call:
Well ok, it's nice to know about what the actions to date have done to lower dilution. But how about the future? What would retiring each additional tranche do for dilution? The section following deals with short interest but the data is the same for share dilution.
Management would do well to come clean with shareholders and say something like, "We felt the issuance of these convertibles was the right thing to do at the time. We feel they enabled us to buy Elpida. But what was a correct capital decision then has turned into a big mistake now, given our rising stock price. We are taking our medicine and moving aggressively to eliminate all of the remaining converts by [xx] date." But this management team has never been particularly candid or forward leaning on items like their capital structure.
Micron's VP of IR, Kipp Bedard apparently made this comment to SA regular Bruce Burnworth:
From Kipp: "We have over 2 billion of outstanding convertibles and the issue we have to keep our eye on is that for every 2 dollars in stock price increase it's cost us and our shareholders 400 million additional to retire the converts. Minimizing the dilutive effect of these converts is still one of the finance teams higher priorities."
7 Apr, 09:46 PM
It is nice to hear that retiring the converts is one of the higher priorities. It would be nicer to hear a commitment on a date. It would be nicer still if it were factually accurate that there are only $2 billion in converts outstanding and that $2 in stock price rise costs $400mm. Neither of these are correct.
What effect has it had on the company's short interest? The first chart above, pulled from the 10Q, shows that all the converts could convert into 189 million shares. This is 161% of the 117.6 million of short interest as of 3/31/2014. What? That can't be. Just shows that not all of the converts are hedged (don't tell the holders they are leaving money on the table!) While these shares aren't those of a "vulture fund" or classic short seller betting on Micron's share price going down, there could be a short squeeze as these shares are repurchased to cover shorts; this is one important reason, besides just the dilution, that Micron should repurchase these converts for cash.
Is there anything that can be done about it? The most recent 10Q indicated that tranches 2031B, 2032C, 2032D, 2033E, and 2033F are currently callable. See note 2 of debt table on page 12 of the 10Q. These should be called and retired for cash right away.
The company should also be aggressively buying, for cash, the non-currently-callable 2043G tranche in negotiated transactions or in the markets NOW and paying more than the intrinsic underlying "as converted" price, and perhaps significantly more. Assuming all the bonds can't be retired through purchases, there is one other technique which has worked in similar circumstances.
Keeping it simple for Ron Foster, he should look at the website "defeasewithease.com" where they have this "cocktail conversation" definition of defeasance:
Put simply, defeasance is a substitution of collateral. Typically, a defeasance is coordinated to close contemporaneously with a sale or refinance. The borrower uses proceeds from the sale or refinance to purchase a portfolio of U.S. government securities that is sufficient to make all of the remaining loan payments. The securities are pledged to the lender, and the lender releases the real estate from the lien of the mortgage. The promissory note, which remains outstanding after the defeasance, and the portfolio of securities are assigned by the borrower to a successor borrower that makes ongoing debt service payments.
I suggest that the company defease the $1099 million of market value, for the $1025 million face value of bonds in the 2043G tranche, RIGHT NOW! The company would deposit $1099 million in US Treasury securities with the bond indenture trustee. To the extent that the treasuries didn't cover the 3% interest for the next 29 years until maturity, the company would true up the deposited amount. Then, it gets a little tricky and some fancy law firms are going to earn a lot of money. The company doesn't pay the next interest payment in order to force the bonds to default and thus to be immediately due and payable. The fancy law firms opine that this default on a defeased bond does not cause a larger technical default and mistakenly bankrupt the company. This last step, on defaulting on the bond, is necessary to stop the clock on the "optionality" of the underlying conversion feature, since the bond holders no doubt want to hold the bond until the first "call date" on November 20, 2018 if not all the way to maturity in 2043. Defeasance isn't pretty and may result in lawsuits. It's been done plenty of times and ought to work here. Does this sound like high risk? Maybe, but so is letting this tranche of bonds grow by $35 million in intrinsic value for every $1 stock price rise between now and the first call date.
Conclusions. Micron has plenty of cash flow and plenty of cash and should retire these convertible bonds NOW and minimize dilution. Even after the 2014 maturities are retired, apparently during this current third quarter, the company is still exposed to about a $189 million increase in the intrinsic value of the convertible bonds for each $1 rise in share price.
Ron Foster is truly the multibillion dollar man. He spent $250 million of the company's cash on poorly constructed hedges on the Elpida transaction. He's spent hundreds of millions on investment banking fees for issuance, redemption, and exchange of these disastrous convertibles. He's spent $194 million on a very ineffective capped call transaction. And he has spent several billion on the principal increase, cumulatively, for these various tranches of bonds. Ron Foster should go. The board should be scolded. Are you listening Baupost and Greenlight?
The Micron owning hedge fund partner (Mr. "Japanese Garden"), who looked at a draft of this article, had this to say, "I think you should add that issuing converts to save a few basis points of coupon at a time when interest rates are lower than at any point in 400 years makes little sense." I can't attest to his interest rate assumptions over the last 400 years, but I agree that the spread between the convert coupons we issued and the straight debt we might have issued, didn't make this convertible saga a sensible idea in the first place. I suspect that soon these passive hedge funds won't be making these comments to an inked stained wretch on Seeking Alpha, they will be shouting it forcefully on the floor of the annual meeting and in proxy fights.
Memo to Ron and the board: Straight debt, no converts.
Disclosure: I am long MU. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.