Open Market Operations And Statistics

Tyler Durden's picture

Zero Hedge stirred up a hornet's nest recently by bringing to popular attention Chris Martenson's post highlighting the very rapid "uptake" via Permanent Open Market Operations of a Treasury CUSIP that had been auctioned off shortly prior. The article angered among others such bloggers as former employee of the NY Fed John Jansen who penned a post titled "Monetizing the Debt: Disinformation in the Blogosphere" whose conclusion was that despite the 5 day turnaround to monetization of the particular 7-year CUSIP which represented roughly 47% of the total purchased by primary dealers "there is absolutely nothing unique or special about today’s transaction by the Open Market Desk." Subsequently Jansen proceeds to nitpick semantics but we will let that slide as it is not relevant for the time being.

Yet what neither Jansen, nor Yves Smith, who picks up the baton and calls Zero Hedge's style "hyperventilatory" (we most certainly do not mind - our style such as it is, hyperventilating or otherwise, at least brings broad attention to topics which may or may not be of relevance to the general public, as opposed to a closed group of highly sophisticated economists and financiers who enjoy debating among each other and perpetuating their closed group relationship, with no informational leakage into the broad arena - a theme that has persisted for many years and is significantly counterproductive to the ambition of bloggers to be an alternative venue to mainstream media), focus on, and what was the primary argument of my, and Chris Martenson's post, is that Ben Bernanke is essentially monetizing debt, despite potentially perjurious claims to the contrary, all within the legal framework of primary deal intermediation and the disadvantaging of indirect dealers (who are very hard to be seen at the table these days as is). Neither Jansen nor Smith point out the fact that a primary dealer, whoever it may have been, would purchase CUSIP 91282LD0 on July 30, and then sell it at a loss less than a working week later (purchased at 99-26 and sold at 99-07). Absent a backstop from another entity it would seem a rather  imprudent thing to do from a fiduciary point of view, especially with such a brief turnaround timeframe: primary dealers have the balance sheets to be patient when they acquire Treasuries.

All that being said, Zero Hedge decided to perform a statistical analysis of time series variance between Treasury auctions conducted in 2009 and the subsequent Permanent Open Market Purchases, in which we analyzed the size of the OMO transaction, the number of days passed between a certain CUSIP being auctioned and becoming purchased by the Fed, the percentage of primary dealer purchases as a total of the entire Competitively Accepted amount, and the percentage of the Primary allocation that would end up being purchased subsequently by the Fed in the OMO-to-Auction timeframe. The results were surprising.

But first, I will present the data set that Zero Hedge created with the assistance of reader Phaesed. The chart below highlights all Bond (not Bill) Treasury Auctions conducted in 2009, segregated by auction date, and highlighted by tenor.

Some datapoints: there has been $1.1 trillion in Bonds offered YTD, of which $1,156 billion has been accepted. Of this total, $600 billion has gone to primary dealers. The primary allocation has represented a (simple) average of 55% of the total bonds allocated competitively. For readers who would like to play around with this data, it will be posted shortly to google spreadsheets for open source enjoyment.

Next, Zero Hedge compiled all the YTD Open Market Operations Data, and filtered it by Bond issues auctioned in 2009, and subsequently purchased by the POMO program.The raw data is presented below:

Some facts: the Fed has purchased roughly $240 billion in OMO since QE was announced in March: on par to purchase $300 billion as the program was initially intended to expire by the end of September: according to media reports as of now there will likely not be an extension due to the "improving economy." (We have not focused on agencies in this report: we will perform that analysis at a subsequent date, yet one can argue the vast majority of Fed buybacks has occurred in the MBS realm)

Of this $240 billion, almost half, or $112 billion has been targeted at treasury issues auctioned off in 2009. The chart below demonstrates the global universe of all OMO purchases of 2009 issues indexed by total purchase size, as well as number of days of OMO transaction since original auction (horizontal axis) - this is the key topic in question, which both John and Yves seem to take offense to. Of course, while a normal distribution would not show a significant preference to a lumped clustering, the vast majority of the treasuries purchased by the Fed has been within a few weeks at most of the original Treasury auction.

The size of the various circles is a relative indication of the percentage the OMO purchase representated as a function of the original primary dealer allocation: potentially an indication of how unwilling the primary dealer may have been to purchase a given issue absent a backstop guarantee from the Fed that shortly after the auction the issue would be acquired via OMO. 

As the chart is somewhat noisy on the tail end, we have cleaned it up once by removing any OMOs that were less than $1 billion in size. The data is as follows:

What becomes notable is the mentioned clustering around the proximal side of auction-to-OMO time, with greater amounts purchased by the FED for CUSIPs that had large primary dealer allocation.

Last we perform one final filtration of the data, to remove any issues that were repurchased by the Fed more than 30 days after the Treasury auction:

The startling conclusion: $32 billion of Treasury Bonds spread across 7 CUSIPs, were purchased by the FED within 10 days of their initial auction and allocation to primary dealers. The amount purchased by OMOs represents an average of 32.4% of the total allocated to primary dealers in the respective auctions. Furthermore, almost two thirds of total OMO Operations for bonds issued in 2009, or $62 billion, affects Bonds issued within 30 days of the OMO purchase. These purchases account for a total average of 29% of the total amount allocated to primary dealers. While one may make the argument that on the run bonds are preferred on average by the Fed for purchasing and by the primary dealer community for selling, the data presents a marked skew in the Fed's desire to monetize very recently issued Treasuries.

The key questions remain: allocations to primary dealers in 2009 Bond auctions is an undisputed majority (55%) of all auctions - this is troubling due to the the recent change in the definition of indirect purchasers as well as the markedly reduced interest of foreign buyers such as China and other indirects, for US Treasuries. Could a reason for the Chinese lack of appetite be due to the fact that while primary dealers represent not just a majority of all Treasury purchases, that these dealers may also have an implicit understanding that come hell or high water for auctions that lack indirect interest, the Fed could potentially make any dealers whole on purchases and subsequent sales at a loss such as the highlighted CUSIP 91282LD0 example (explicitly, at a loss for taxpayers who have to fund the primary dealers shortfall, in this case the difference between 99-26 and 99-07)? Would the Chinese be interested in playing in a rigged playing field when indirects are potentially impaired vis-a-vis direct purchasers? Furthermore, is Bernanke pulling a Clinton and while claiming under oath the he is not monetizing debt, he is effectively doing just that on well over $30 billion in Treasuries, which the Fed acquires within 10 days of issuance? And lastly, is the rapid uptake by the Fed a means to goose up auctions which have a potential likelihood of failure: the 7 Year in question came hot on the heels of a 5 Year that for all intents and purposes was quite close to a failed auction? Absent an implicit backstop, which everyone knows the Fed is very keen on making these days: as the SigTarp demonstrated, to the tune of tens of trillions of dollars, what is the likelihood the 7 Year would have fared as well as it did, had not the primary dealers really stepped up, for reasons known and unknown.

Zero Hedge is not making any claims, but merely asking questions. And while we appreciate the opinions of self-professed experts such as John Jansen, these answers should really come from the proper authorities - the US Treasury and the Federal Reserve of the US.

As time allows, Zero Hedge will next conduct a comparable study on Agency and MBS debt repurchases by the Federeal Reserve.