By Tom Adams, an attorney and former monoline executive; Andrew Dittmer, a mathematician who has worked for a hedge fund; Richard Smith, a UK-based capital markets IT consultant, and Yves Smith
As described in ECONNED and in later reports by ProPublica, a Chicago-based hedge fund, Magnetar Capital, entered into a program of sponsoring subprime-based CDOs that was unprecedented in its scale. The hedge fund supplied a disingenuous, legalistic defense of its strategy, in an effort to depict the ProPublica reports as having failed to integrate some of the information Magnetar claims to have provided to ProPublica.
However, a close reading of their letter shows that it does not effectively rebut the overarching charge, that Magnetar-sponsored CDOs were designed to fail, but is also consistent with our analysis of the strategy (which came to Magnetar’s attention more than three weeks ago).
Magnetar’s efforts to defend itself consist of some seemingly straightforward arguments, as well as two dense paragraphs. We’ll deal with the more accessible bits first:
1. Its strategy was “market neutral” and would show a profit whether the subprime market did well or badly
2. Magnetar was merely the “initial equity purchaser”; the “Collateral Manager” (often know as the CDO manager) and the dealer were independent, and the Collateral Manager was responsible for selecting the assets and exposures in the CDO and had a financial incentive for the deals to succeed
3. Magnetar’s analysis of its CDO’s performance, contra that of ProPublica, shows that its CDOs performed better than that of “like securities”.
Let us debunk each in order.
1. merely means the trade was profitable whether the underlying market went up or down, not how the profit potential was distributed across the two outcomes. We have indicated, based on the input of market participants, including individuals who worked on Magnetar’s deals, that the ABS (subprime-related) CDO trade was constructed to show a comparatively thin profit if the CDOs continued to perform, and a much greater return if the bonds in or referenced by the CDO failed. This bias is consistent with the 76% returns for this strategy reported by ProPublica.
2. is simply disingenuous. Substantial equity investors were given considerable rights to influence the overall design of a transaction, with the notion that any steps the equity investor took to protect the value of his investment would benefit all other investors. There is ample evidence that Magnetar used the influence it gained over transaction to influence its parameters to its benefit, to the detriment of the other investors/funders on the long side. For instance, it used triggerless deals (in a typical CDO, the cash flow distribution to the equity trance would be cut when losses reached certain thresholds. In a triggerless deal, the equity tranche investor would have the same rights to payment right up to when the CDO defaulted or was liquidated.
Some accounts indicate that Magnetar provided lists of suggested securities to the CDO manager, it would not have been necessary to go that far to influence CDO design. For instance, merely calling for a very high average coupon would have forced the CDO manager to buy only particularly “spready” or high yield, meaning crappy, bonds. The ProPublica article includes an e-mail message from Magnetar’s James Prusko to a CDO manager, Ischus, where Magnetar not only pushed for higher spread (meaning riskier) CDS on subprime bonds to be included, but also provided a spreadsheet with a “target portfolio”. Magnetar would achieve its ends if the CDO manager used many of the instruments its suggested portfolio, or simply constructed one that had the same characteristics.
While we have taken issue with some elements of Michael Lewis’ The Big Short, his book does provide a good description of the role the CDO manager played: that of creating the appearance of independent, objective asset selection. In reality, many CDO managers were closely affiliated with particular investment banks and were so thinly staffed that the idea that they were actually doing much analysis is questionable. CDO manager fees were 0.10% to 0.20%, meaning $1 to $2 million per annum on a $1 billion CDO for doing very little.
3 is ridiculous. Moat of the deals have hit an event of default and the average rating is well into the CC range. The deals all failed,so whether they failed in 10 months or 12 months is not a measure of success. The weighted average Moody’s rating of senior bond of magnetar deals is Ca (CC equivalent). 64% of the deals had their ratings withdrawn. Arguing that their deals performed better than some other deal is meaningless semantics.
Now we will parse the denser section of their argument, with Magnetar’s text in italics:
Magnetar would not have invested the way it did if it had concluded that the housing and residential mortgage markets were near collapse. If Magnetar had that view, it would not have needed to devote resources to its strategy of combining long positions with hedges. Short positions were easy to obtain.
True, but a misdirection. Note the word “near” in “near collapse.” Short sellers almost never are able to short a bubble right before its peak. Magnetar’s strategy was designed to address the usual problem that vexes short sellers, that they often wind up being early, and conventional strategies for shorting show losses until the market turns. Magnetar’s strategy was designed to avoid the problems other shorts had had who had underestimated how long it would take for the bubble to pop.
Magnetar’s strategy for investing in CDOs was based on a market neutral mathematical statistical model, and was designed to have a positive return whether housing performed well or poorly.
This is consistent with what ECONNED and ProPublica have written. This statement does not mean “Magnetar’s strategy would have a positive return no matter what,” since no such strategy exists. It is a labored way of saying “there were scenarios in which housing performed well under which Magnetar’s strategy would be profitable, and there were also scenarios involving a housing collapse under which Magnetar’s strategy would show a profit.” Most likely, all of this just means that Magnetar was long
correlation and that the return from their equity position in the CDOs was fat enough to fund
their short positions.
Magnetar’s statistical models looked at all equity tranches and all hedges in the portfolio simultaneously in a global framework and not on a deal by deal basis.
Again, this is not inconsistent with what we have written. None of this contradicts the thesis that Magnetar knew very well that certain types of CDOs were better for their strategy than others, and actively pushed to make sure those sorts of CDOs were more likely to be created.
Magnetar explained this market neutral strategy to its investors beginning in early 2006. Magnetar’s strategy was not based on fundamental analysis expressing any view that values in the housing market would go up or down.
The catch here is “fundamental analysis..of the housing market.” The insight that drove this arb was that risky subprime loans looked pretty certain to come to a bad end, and that the price of insurance on this credit looked to be wildly underpriced. You didn’t have to believe the housing market would decline to think this trade was a good bet.
Magnetar’s model instead focused on the structures of the mortgage securities and the CDO markets.
…. i.e. the fact that CDO assets were massively correlated, which meant that the AAA tranche of a CDO was badly mispriced.
Magnetar employed no fundamental analysts to make its investments in CDOs.
As of the end of September, 2007, the majority of the notional value of Magnetar’s hedges referenced CDOs in which Magnetar had no long investment.
This statement is tricky. The Wall Street Journal has noted that Magnetar took short positions in addition to the shorts it constructed using the CDOs it sponsored. One interpretation is that Magnetar is referring to CDS written by third parties against CDO tranches, meaning those cases where Magnetar bought protection on a CDO tranche from a protection seller of some sort. These transactions were unusual, and would be likely to be only a minor component of Magnetar’s overall position. It could also mean is that Magnetar bought CDS against CDOs containing the same sub bonds that appeared (in synthetic form)
in the CDOs that they sponsored.
Notably, focusing solely on the group of CDOs in which Magnetar was the initial purchaser of the equity, Magnetar had a net long notional position.
Magnetar is trying to give the impression it did not have a net short position on the CDOs it sponsored, but the writing awfully tortured. For instance, “initial purchaser of the equity” (as opposed to more straightforward terminology like “sponsor”) may narrow the universe under discussion to a small subset of the trades we and ProPublica have identified. It might even mean simply that their short had positive carry.
To put this into perspective, Magnetar would earn materially more money if these CDOs in aggregate performed well than if these CDOs performed poorly.
This statement appears to be broad (as in arguing that Magnetar would do better if the CDOs it sponsored did well) when it could be very narrow. For instance, we are now discussing “these CDOs”, which per above may be a subset of all the CDOs Magnetar created. Moreover, our analysis suggests that Magnetar bought only a small percentage of the CDS protection created by its CDOs (we assumed a short interest 4x its long position, which would be equivalent to roughly 20% of the par value of the CDO). With this structure, Magnetar does best in the unlikely event that ONLY the bonds referenced by its CDS fail, and the rest perform. It would not only collect on its short bet, but the CDO would only be somewhat impaired rather than fail.
The purchase of credit protection for Magnetar was primarily a portfolio hedge to Magnetar’s long positions. This distinguishes Magnetar from some other market participants that purchased credit protection primarily as an expression of a fundamental view on the market and not as a hedge.
If our understanding of the Constellation program is accurate, this statement may be narrowly true, but it is highly disingenuous. The implied opposition between “portfolio hedge” and a short position on the market probably refers to the contrast between the Magnetar program that was long correlation and people like John Paulson and Steve Eisman who were simply short housing.
Magnetar’s role as an equity investor was limited and did not replace the roles of the investment bank (Dealer), the Collateral Manager or the credit rating agencies.
By design. But none of this contradicts any of the things that are alleged about Magnetar’s role as an equity investor.
Collateral Managers and Dealers were independent from Magnetar and negotiated at arm’s length against Magnetar with respect to transaction terms, such as the amount of certain types of compensation that the Collateral Manager and the Dealer would receive from the CDO. Magnetar did not select the underlying assets of the CDO at any time prior to or subsequent to transaction issuance.
Doesn’t follow that they were not successful in influencing the deals’ parameters.
When Magnetar was an initial purchaser of the equity of a CDO, the compensation terms and structure of the transaction provided the Collateral Manager with a financial incentive to select assets they believed would perform well.
This remark implies that the system of using Collateral Managers who received incentive compensation helped create high quality CDOs. The results across the industry in fact were the polar opposite.
Before Magnetar’s participation as an equity investor, the fee arrangements generally had less emphasis on incentive-based fees.
If we assume, then, that Magnetar is suggesting that Magnetar favored compensation packages for managers that were heavy on “performance incentives” and light on direct compensation, one tempting explanation is that they figured that the performance incentives would be unlikely to be paid.
In those CDOs in which Magnetar was an initial purchaser of equity, Magnetar did not have the intent or any reasonable basis to believe that the CDOs were built to fail.
This is the hardest statement to make sense out of, particularly since it does not track grammatically. But the “built to fail” by implication refers to the action of third parties (since Magnetar is insistent that it its role in influencing the deals was very limited). So this statement could be taken to mean that Magnetar did not believe that the Collateral Manager or Dealer were designing the CDO to fail.
In addition, derivatives expert Satyajit Das, who reviewed the Magnetar letter, advised us, “Because of model differences and problems of verifying the inputs, it is possible to construct speculative positives that may superficially appear hedged. This allows traders to take any position that they really wish to.”
This means that Magnetar could muster a mathematical analysis of its portfolio that would support its claim that it was hedging, when in fact its intention was to take a short position.
Importantly, the study finds that, when the same Collateral Manager executed transactions with and without Magnetar as an equity investor, the transactions in which Magnetar was an investor performed materially better using the above key metrics.
ProPublica claims to have produced a study showing that Magnetar-sponsored CDOs performed worse than industry norms, which is consistent with our own review of the transactions. All of
the Magnetar CDOs blew up, so the point is somewhat academic.
Put simply, Magnetar’s defense, while predictable, is less than persuasive.