A nice thing about IPOs is that they end: you work for months on pitching and executing a deal, you write hundreds of pages of documents, you embark on a roadshow with tiny planes and slovenly CEOs, you price the deal, you watch it trade on that first exciting day, and then you don't do it any more. You and the company bask in the warm glow of a successful deal, and/or you avoid their phone calls in embarrassment about a bad deal, and then you give them some space before coming back and pitching them on the next piece of business.
Facebook is like the opposite: no bank involved in it can be in that much of a hurry to pitch the next piece of business, but the IPO itself will be relived over and over again for the rest of time. The latest is Citi's angry letter to the SEC, responding to Nasdaq's proposal to compensate market makers who lost money on the deal. Their preference is, uncharacteristically, to be compensated more, and they express that preference in the form of a litany of complaints about Nasdaq's ineptitude and self-interest.
Furthermore, while market participants were hemorrhaging money, Nasdaq made $10.8 million from trading out of a position in Facebook shares that it had taken during the commotion on May 18. Nasdaq's explanation is that "[c]ancellations received [between 11:11:00 and 11:30:09 AM] were processed in real time, resulting in Nasdaq assuming in its error account the cancelled buy and sell positions." Notably here is the fact that Nasdaq market makers were unable to do what Nasdaq had done. Market makers could not see their positions in real time, nor could their customers.
So, that does actually sound pretty unfair: Nasdaq got to fix its erroneous positions in near-real-time, but nobody else could find out their erroneous positions until much later. A while back we talked about this error account position and I guessed that Nasdaq was accidentally short about 3mm imaginary shares that it then had to buy back, effectively propping up the Facebook stock price. So that seems to have been right - Nasdaq counts 3,070,430 imaginary short shares. I imagined Nasdaq putting that imaginary short on at $42 in the opening cross, and then covering it the afternoon of the IPO at a price a little above $38, which seems to have been about right. And I speculated that that Nasdaq's covering of its short helped - a bit - to prop up the price on that Friday.
But Citi's letter suggests another, bigger thing that helped prop up the price. Here is Citi arguing that it should be compensated for phantom shares based on the Monday-morning VWAP rather than the Friday-afternoon price, because nobody could figure out how many shares they had until the weekend so any dumping of erroneous positions would have to occur on Monday:
At a minimum, Citi believes that the appropriate level of compensation should be measured using a volume weighted average price (VWAP) calculation for shares of Facebook between (a) the opening price on Monday, May 21, and (b) 12:00PM on Monday, May 21 .... The justification for this VWAP measurement is that, while Nasdaq purports to believe all member firms should have known their positions at exactly 1:50PM on Friday, May 18 (when Nasdaq first began sending back to its customers execution reports for the IPO Cross that occurred at 11:30AM), the reality is that the vast majority of Nasdaq's customer base did not know their true position until well after the close of business on Friday, May 18. ... Thus, Nasdaq knew that the majority of its customers were forced to carry significant positions over the weekend, and that they were not able to close those positions until trading began on Monday, May 21.
Here is a simplified description of the Facebook IPO mechanics:
- The underwriters priced and sized the deal aggressively, increasing the likelihood that it would fall through the $38 IPO price.
- On Friday, the first day of trading, the deal did in fact look likely to break through that price, forcing the underwriters to buy in ~2/3rds of their greenshoe - some $1.5bn worth of stock - at $38ish.
- There was also a bit of other support - 3mm shares, or $120mm-ish worth of stock - from Nasdaq buying in its phantom short on Friday.
- More notably, though, some chumps were stuck with billions of dollars of stock through the weekend - because they didn't know they had it until late on Friday afternoon and couldn't sell until Monday morning.
- On Monday morning, after everyone had figured out what they had and decided to sell their unwanted shares - the stock opened at $36.53.
- For Citi, which was left long stock over the weekend, this was unpleasant - which is why they want to get compensated based on the Monday morning price.
- For others, it was even more unpleasant: UBS, for instance, seems to have accidentally gotten long several billion dollars worth of stock on Friday, and it's unclear they were able to get out of all of it before Monday morning. Others - Knight Capital, etc. - were in the same boat. Presumably some of them dumped their erroneous shares on Friday afternoon; others, like Citi, did not.
- For certain other people this was highly pleasant. Most notably, the underwriters, who would have probably eaten through their entire greenshoe (and then some?) if all of those shares had come for sale on Friday, instead got to wait until the IPO broke through its deal price in Monday's opening cross. So they ended up buying in 1/3 of their greenshoe, some $800mm or so worth of stock, at $35ish on Monday, after giving up defending the $38 price, rather than having to buy those shares at $38ish on Friday afternoon when the deal was still defensible. This is how they made $100mm.
It's not obvious how you put numbers to that, but for a ridiculous rough-and-ready calculation, consider that 107.5mm shares traded between 9:30 and noon (Citi's proposed VWAP period) on Monday at a $34.6001 VWAP, versus 60.4mm shares during the same period on Tuesday. Assume that half of the 47mm "excess" shares sold on Monday morning would have been sold on Friday had their owners known they owned them and you get 23mm shares,* or very roughly the amount of greenshoe shares that the underwriters did not have to buy in on Friday afternoon - and could instead buy in at much lower prices on Monday morning.
In other words, Nasdaq's errors very plausibly caused the entirety of the underwriters' trading profits.
Now, no step of the above is particularly nefarious, but they combined to create a wealth transfer from market-making chumps like UBS and Citi to underwriters like Morgan Stanley and Goldman Sachs. Oh, and Citi. DealBook reports that Citi lost $20mm on Nasdaq's errors and wants it back. They should at least credit the $2mm or so that they made on the greenshoe - 2.25% of those $100mm gains - against those losses.
Citigroup Assails Nasdaq Over Flawed Facebook I.P.O. [DealBook and pdf letter]
* Also, if you go along in that assumption, I have a punch in the face that you might be interested in. Still, surely Citi is right that some, probably significant, chunk of shares coming loose on Monday would have come loose on Friday had their owners known about them promptly.