When people think failed or busted Treasury bond auction, they usually imagine something out of Brazil or Russia where the government was selling obligations and nobody showed up. Of course, in the US, courtesy of the Primary Dealer system and more importantly, of a multi-trillion shadow banking system, where bonds are cash equivalent following rehypothecation and pledging for cash-equivalents with virtually no haircut, there is no risk of an auction failing in the conventional sense, at least not until Bernanke finally manages to irrevocably erode the Dollar's reserve currency status. However, that does not mean that auction's can't "fail" in a purely technical sense. Which is exactly what happened during last week's sale of 3 month Bills, when due to a "glitch" in the system not only was a key Primary Dealer locked out of the auction, forcing the US Treasury to arbitrarily reassign allotment in the parallel 6 month auction, but leading to a wild intraday mispricing in the already collateral-scarce, short term maturity market.
The dealer in question is Wall Street's biggest pure-play hedge fund (excluding JPM's CIO office, and the Federal Reserve of course) Goldman Sachs, whose order at the September 9 3-Month bill auction did not come through, making it seem that there was far less demand for the paper, and resulting in a brief spike in the yield of 3 Month bills, which priced at a 0.02% yield, or about 50% higher than the 0.013% yield for the complex before the auction.
The Treasury, meanwhile, scrambling to remedy the situation, had no choice but to boost allocation to Goldman in the concurrent 6 Month Bill auction. As WSJ reports, "The Treasury made the decision to give Goldman more six-month T-bills than it had bid for in the two-minute span between the auction ending and the results being disseminated to the market. In making this decision, the Treasury chose to break one of its own rules for its debt auctions. The Treasury limits the amount of debt any one buyer can obtain in an auction to 35 %.... [as a result] the six-month bill auction appeared to have more demand and the bills sold at a lower rate than expected, at 0.035%. That is a tumble from comparable bills yielding 0.048% before the auction. Six-month bill yields fell to as low as 0.02% after the results were released. "
Sadly, the Treasury has no problem with allowing the Fed to monetize up to 70% of any one CUSIP, albeit it happens in the secondary market from the same Primary Dealers who buy paper outright. So it is not technically "monetization."
This is the announcement that the Treasury posted on the day after the auction:
During yesterday's auctions of 3- and 6-month Treasury bills, the Bureau of the Public Debt (BPD) discovered a technical issue within the TAAPS system, a web-based portal for bidders to participate in Treasury auctions, which resulted in one bidder being unable to access the 3-month auction. Our manual attempts to address the issue ultimately resulted in the bidder being awarded a larger sum of 6-month bills than originally intended, in excess of the '35 percent rule' for competitive bidders in Treasury auctions. In this instance, we have determined that it is in the best interest of market participants to waive the 35 percent limitation and for yesterday's auctions to stand. Today's scheduled auctions occurred without issue and BPD continues to test its systems to maintain the smooth functioning of Treasury auctions.
The WSJ correctly observes that the, "incident rattled the short-term debt market at a time when investors in U.S. government debt are adjusting their portfolios amid concern about the path of interest rates."
Luckily this time it was just a Bill auction, which is not as closely scrutinized by the broader public, although the vast majority of money-markets, banks and corporations are very active participants in the cash-equivalent space. However, what if the TAAPS system encountered a "Taps" moment, and shut off not just Goldman but the majority of Primary Dealers (we hope the Syrian Electronic Army isn't getting any ideas here), in a much more important benchmark security, say the 10 Year? How would the already jittery market react if instead of pricing just around the When Issued, the 10 Year were to price some 50% higher for some unknown reason?
One can only imagine the cross asset cataclysm that would ensue as the panic gripped the kneejerky, HFT-dominated bond markets, who, unclear what had just happened, decided to follow suit and dump it all...