There's a juicy pile of something going on over on Maiden Lane. Once upon a time, Goldman Sachs murdered AIG and stuffed its corpse with tons of shall we say "troubled" residential mortgage-backed securities. Like a cursed diamond, those securities then bounced around among owners who came to bad ends and ended up in a thing called "Maiden Lane II," owned by the New York Fed and managed by BlackRock, with a mandate to sell them off over time at prices that "represent good value for the public."
One day, Credit Suisse came to BlackRock with reverse inquiry for those Maiden Lane II bonds. The Fed via BlackRock solicited bids from five banks, CS, Goldman, Barclays, RBS and Morgan Stanley. The banks conducted some pre-bidding price discovery with their clients, though they were sworn to secrecy and had to get the clients to sign nondisclosure agreements before they could solicit them. Eventually the banks put in bids and Goldman won and bought the bonds, and is now selling them rather nonchalantly to clients, keeping most of them overnight after buying them from the Fed.
A simple story, but it raises two interlinked things to worry about:
(1) Why are you giving all those wonderful wonderful bonds to Goldman, huh NY Fed? HUH?
(2) Why are you keeping all those deadly deadly bonds on your balance sheet, huh Goldman? HUH?
For the first one, Bloomberg says:
After inviting more than 40 broker-dealers to take part in a series of auctions last year, the Federal Reserve Bank of New York asked only Goldman Sachs Group Inc., Credit Suisse Group AG and Barclays Plc to bid on the full $13.2 billion of bonds offered in two sales over the past month.* The central bank switched to a less open process after traders blamed the regular, more public disposals for damaging prices in 2011. This week, Goldman Sachs bought $6.2 billion of bonds in an auction.
The selectivity has irked firms that weren’t also given the chance to profit from the auctions, and raises the question of whether the Fed got the highest price for U.S. taxpayers, who gave insurer AIG a $182.3 billion bailout. The New York Fed resumed its sales of the assets in January after the market recouped a portion of last year’s losses.
Now I personally am unsurprised that people who didn't get to bid are sad that they didn't get to bid since that was apparently a way to make money and people like money:
Goldman Sachs told some investors who bid on the bonds through the bank that, while they had offered the best prices on individual securities, the firm had bought the debt for itself, according to three money managers with knowledge of the matter. Goldman Sachs then offered the securities for sale to the investors, they said. The prices were between 1 and 3 cents on the dollar higher, said one of the people, who declined to be identified because the transactions aren’t public.
Got that? The Fed contacted the banks and asked them to bid. The banks could contact clients, subject to nondisclosure agreements. Some clients signed those and bid, some didn't and didn't. But Goldman wasn't bound to honor any bids - they were basically just indications of interest so Goldman could formulate its own bid. It then put in an at-risk bid, took all the bonds, and went back to those who expressed interest and offered them the bonds at 1 to 3 points above what Goldman had paid (and above what the clients had bid).
WTF, you say, what a bunch of scumbags. Well keep in mind that this behavior would make no sense if the only possible buyers were the original NDA-bound clients. Why would they raise their bid just because Goldman, rather than Maiden Lane, was the seller? The answer has to be that Goldman could create additional price competition by selling more broadly - once the sale was announced by the Fed, Goldman can sell to whomever they want with no NDAs.
So in fact Goldman didn't mostly get the things off their books instantly. Bloomberg and Reuters went and did some clever sleuthing in the TRACE data and found that on February 8 dealers of non-investment grade MBS had $6.9 billion in "customer sell" tickets (with Maiden Lane II presumably a customer) and only $960mm in "customer buy" tickets,** meaning that GS had sold very little of its exposure as of the close of business on the day of the auction. As Bloomberg puts it:
“It certainly looks like they’re willing to provide liquidity to the market by bidding for their own account,” said Guy LeBas, chief fixed-income strategist at Janney Montgomery Scott LLC in Philadelphia. “It’s a little bit of shift,” after a period in which dealers reduced their inventories, he said. ...
“Our intention has always been to distribute the portfolio to our clients globally and we are in the midst of doing that,” Michael DuVally, a spokesman for New York-based Goldman Sachs, said today in an e-mail.
Now, buying $6.2bn of mortgages prop would be rather outsized for GS, which had a little under $8bn of total residential mortgage exposure last time anyone checked in September 2011:
"If the Volcker rule had been in place, would this have run up against it? That is exactly what is being debated right now. Are trades like this going to be viewed as consistent with the proposed implementation of the Volcker rule or not?" said Darrell Duffie, a finance professor at Stanford University.
I think the better answer is "or not." The Volcker Rule proposal says:
... bona fide market making-related activity may include taking positions in securities in anticipation of customer demand, so long as any anticipatory buying or selling activity is reasonable and related to clear, demonstrable trading interest of clients, customers, or counterparties.
Which this more or less seems to be - as Reuters notes, "Each of the brokers, in bidding for the portfolio of bonds, lined up customer bids for the bonds. In an indication of the demand, one bank received more than 1,000 bids." But it's a legitimate question - if you're worried about banks putting tons of their own money at risk on terrifying mortgage-backed securities, a bank putting $6.2 billion of its own money at risk on the very mortgage-backed securities that killed AIG is the exact thing you were worried about. Maybe even if that money is just at risk for a few days.
So what to think of all this? Two things come to mind. One is that Steven Davidoff has a good article in DealBook today to the effect of "the underwriters of the Facebook IPO can make you fools pay any number they want for Facebook shares." That is maybe a touch exaggerated but you should not forget the basic message, which is:
(1) Morgan Stanley's job is to get investors to pay lots of money for the securities it's selling,
(2) Morgan Stanley has various tools with which to accomplish that job, and
(3) Morgan Stanley is good at its job.
Facebook, if it cares about the valuation of its IPO (which I guess it should? meh), should hire someone who is good at creating demand. Similarly, the Fed, if it cares about getting a price that "represents good value for the public," wants the top handful of dealers to be involved - and it wants them to follow the procedure that will get them to the best price. The people who were not involved are happy to tell Bloomberg “I fail to see how running a limited participation, secret auction is any way beneficial to the owners of these bonds, the U.S. taxpayer," but like tell that to Mark Zuckerberg, y'know? Oceans of ink are spilled on whether book-building or auctions are better ways to sell securities, but some smart people seem to think that book-building works - that's why it keeps being the method of choice for IPOs. And for Maiden Lane, it seems to have worked: because GS can reoffer at a higher price than its quasi-auction-bidder clients bid, that suggests it could pay the Fed more than a public sale direct to customers would have raised.
Mr Viniar suggested that the rule might lead to banks buying assets and then selling them to their customers at a higher price, in an effort to compensate for the added risk of holding the positions. A bank like Goldman Sachs could then pocket the difference from the wider ‘bid-ask’ spreads, Mr Viniar intimated.
“Ultimately, it could lead to lower dealer inventory levels and could be [return-on-equity] enhancing as we adapt to a less capital intensive business model,” Mr Viniar said in prepared remarks.
This is sort of that, right? Part of the reason that the bidders are limited is that the Fed was going with dealers who it thought could quickly and confidentially offload these bonds. Because that would get it a better price. But "could quickly offload bonds" is a key determinant in the Volcker Rule: you can take down $6bn of RMBS if you have clear, demonstrable customer demand for them - and sell them at a 1-3 point markup. But you can't do that if you don't have that demand - so you never get to see that markup.
My guess is that the Volcker Rule probably will lead to lower dealer inventory levels, wider spreads, and other evils that banks and somecustomers worry about. But the biggest dealers with the best distribution networks will have the easiest time claiming compliance with the Volcker Rule - "sure we bought $6bn of RMBS at risk, but it's because we expect that much customer demand." The smaller guys who were sad to be shut out of the Maiden Lane auction will have a hard time making that argument - and so won't be able to muscle in just by undercutting price. Which will mean less pressure on pricing for the big dealers who can have the field to themselves.
New York Fed Sells $6.2 Billion in Face Amount of Maiden Lane II LLC Assets; New York Fed Loan to be repaid in full [NY Fed]
Fed Plays Wall Street Favorites in Secret Bond Deals: Mortgages [Bloomberg]
Goldman subprime bet highlights Volcker debate [Reuters]
Goldman Sachs Retained Almost All of Fed’s Mortgage Bonds [Bloomberg]
A $100 Billion Value for Facebook? That May Be Possible [DealBook]
Goldman executive talks up Volcker rule [FT]
** Here in all its shininess, the TRACE data for non-investment grade collateralized mortgage obligations traded on February 8: