Negative Specials Rates
The term “Specials” in the Repo market refers to specific securities, like the current U.S. Treasury 10 year note, which trade at lower than regular market rates. At times, short-demand for a security can outweigh the available supply causing settlements to fail, which can then lead to negative Repo rates.
Before 2003, negative rates in the Repo market were quite rare because market participants always had the option to “fail” on the trade (fail to deliver the agreed upon trade). A fail had no costs, or at least, the costs were considered nominal. If someone was unable to cover a short and willing to accept the equivalent of a 0.0% rate, they simply chose to fail. As long as securities shortages were relatively short-lived, there was no reason to borrow at negative rates except under very extreme circumstances.
In the 1990s, if an unexpected securities shortage arose and some bonds were desperately needed, like returning bonds to a clearing bank on reversal time, dealers would create a negative Repo rate through bonds-borrowed style transactions. The counterparty with the hard-to-find securities would loan against an equal amount of general collateral (GC), except the spread between the two rates would be more than the prevailing GC rate, generating the equivalent of a negative rate. Trades had to be booked this way because most trade processing systems could not handle negative rates.
The “big break” for negative Repo rates came in 2003, when the 3.625% 5/13 failed for four months, between August and November.* It took an extreme event like that to illustrate the true cost of fails, or at least, show the costs of fails over a long period of time. There was still no Fail Charge and failing was still the equivalent of 0.0% interest rate, but this time, it was different. Broker-dealers began accumulating a regulatory net capital charge, fails became assets on the balance sheet, labor costs in the back-office increased, and at times, the fails were annoying good customers. One issue that quickly became apparent was that no one could call margin against the fails. After months of trading, market prices had moved considerably and it highlighted the significant amount of credit exposure with aged fails. With a failure-to-deliver and a market decline, the security price was trading lower than the price to be received by the failing customer. If that counterparty defaulted, you’d be forced to sell that security to someone else at a lower price. With the bond market having moved several points, there was a lot of credit exposure accumulating.
In order to help dealers who desperately needed the 3.625% 5/13 to clean up problem fails, the Repo market devised the Guaranteed Delivery (GD) Repo trade. The “seller” was supposed to have the securities in their “box” and would guarantee delivery by the end of the day. The rate on a GD Repo trade was always negative and helped facilitate 3.625% 5/13 settlements, but it had a drawback. There was no penalty for failing a GD Repo trade, the trade was merely cancelled if the securities were not received.
“The Year of CTD Squeezes”
There are a number of U.S. Treasury futures contracts traded on the Chicago Board of Trade (CBOT), including the Five Year Note contact, Ten Year Note contact, and Bond contract to name a few. During the expiration month of the front contract, a formula calculates the specific U.S. Treasury which is most advantageous to delivery to fulfill the “futures” delivery obligation, which is termed the “cheapest-to-deliver” (CTD). The security must be delivered by the last day of the contract month and delivered in the morning. In order to make sure a trader has the securities for the morning delivery, the securities are held in the firm’s own account usually for two days before the delivery date. Holding the securities in the “box” makes sure the securities can be delivered promptly in the morning, because a late deliver to the CBOT is extremely punitive, so as they say, “failure is not an option.”
The first hint of a fundamental change in the Treasury futures market came in December 2004. The CTD for the Ten Year Futures contract was the U.S. Treasury 3.0% 2/09 and two weeks prior to the delivery date, the issue failed for several days, prompting traders to re-examine the two day “boxing” period. The next futures contact, the March 2005 Ten Year Futures contract (3.875% 5/09 CTD) traded at a discount throughout most of March as traders were again worried about fails. By the end of March, fails were not an issue, but the size of the delivery to the CBOT was huge, an total of $11.9 billion in securities were delivered.
It was the next futures contract month, June 2005, where it really “hit the fan.” Beginning at the end of May, the June Ten Year Note futures contact (4.875% 2/12 CTD) collapsed. The basis was trading at an unprecedented discount and the issue was completely gone from the market. Traders who had long basis trades (long the bond, short the futures contact) quickly locked-up their positions in their “box,” more than five weeks before the delivery date. Even with Fed loaning over $2 billion 4.875% 2/12 a day into the market, it was not enough to eliminate the fails and shortage.
At that point, the Repo market developed the Prompt Delivery (PD) Repo trade. The previous GD Repo trade did not fulfill current market conditions because of the lack of a penalty for failing, and potentially not knowing for hours later whether the trade was good or not. A PD Repo trade required the “seller” to deliver the securities within 15 minutes or the trade was cancelled. PD Repo trades began trading at negative rates and they allowed Repo market participants to price the value of clearing a high demand security.
Now, the basis for the Ten Year Note futures contract was able to trade again and pricing roughly the implied PD negative Repo rate. The open interest of the contract reached a high of $19.4 billion (out of a $24 billion issued) and the highest negative rate that ever occurred in the U.S. Repo market traded at -30.0%. Someone was that desperate to get of 4.875% 2/12 and close out their short basis position.
At the end of the month, $14.1 billion in deliveries were made to the Board, indicating the market was forced to “box” over $14 billion during most of June. The deliveries also uncovered they mystery of why the basis was trading as such a discount. PIMCO, the large west coast money manager, had taken delivery of $13.3 billion, almost the entire open interest. As it turns out, PIMCO had attempted to “corner” the basis market and built up a large long futures position, expecting the market would not be able to deliver the full open interest. PIMCO was later charged with market manipulation.
After the PIMCO CTD squeeze, the CBOT issued new position size limits beginning in December 2005 for Treasury futures contacts, including a limit of 50,000 contracts ($5 billion) for Ten Year Note futures and 35,000 contracts ($3.5 billion) for Five Year Note contracts. Since then, there’s been no significant CTD squeezes or shortages and no negative rate trading in CTD issues.
Formal Negative Repo Rate Trading
Later in 2005, the Repo market tweaked the PD Repo trade into a GD Repo trade again, but this time the seller was required to pay a penalty if they failed to deliver. It was a better way of trading negative rates because it created a cost for not completing the trade. The Bond Market Association officially proposed the type of trade in December 2005, calling it a Negative Rate Repo (NRR) trade and it was formally adopted in April 2006.
The NRR served as a good market mechanism for negative rate Repo trading until the financial crisis. In response to the unprecedented amount of fails in September and October 2008, regulators were no longer satisfied with the lingering option to fail Treasury settlements. They put the onerous on the Treasury market to come up with a solution. In May 2009, the Fail Charge was introduced. Depending on the level of overnight rates, there’s up to a 300 basis point charge for a fail which was the equivalent rate of a -3.00% for failing. The charge effectively expanded the interest rate playing field by 300 basis points, where previously the lower end of the Repo trading range was at 0.0%, unless there was some kind of guaranteed delivery trade like we discussed. Since the Fail Charge began, Treasury fails are must less common and Specials often trade with negative rates.
Negative General Collateral Rates
The General Collateral (GC) Repo rate is the interest rate an investor receives who is willing to accept any U.S. Treasury as collateral. The particular U.S. Treasury is at the option of the “seller” and that makes GC an overnight interest rate closely related to the federal funds rate. For GC trades to trade negative, it means the entire U.S. Treasury Repo market is technically “Special.”
When GC has traded negative, it has almost always been on a quarter-end or year-end. For example, on 3/31/09 GC traded as low as -.20%, on 9/30/09 a low of -.25%, on 12/31/09 at -1.80% and on 3/31/2010 a low of -.25%. What’s the reason why an entire market like the U.S. Treasury Repo can trade Special? The answer is one part “Window Dressing” and one part trading error.
During a statement period, like quarter-end or year-end, some market participants “pull” their U.S. Treasurys from the market. These investors don’t want to show leverage or market exposure on their balance sheet and the flow of U.S. Treasurys back into customer portfolios can create temporary shortages. When there’s a securities shortage, there’s bound to be negative Repo rates if the shortage is deep enough.
The other reason for negative GC Repo rates is pure trading error. A dealer committed to delivering U.S. Treasurys to a client for a cash investment or into a Tri-Party account did not cover those Treasurys in the morning. Instead, they waited until later in the afternoon, expecting higher overnight rates, but instead the market moved the opposite direction. They found there wasn’t enough Treasury collateral at the end of the day to fill all the demand and rates dipped into the negatives.
It’s interesting to note, GC has only traded at negative rates late in the day on a statement period. Of course, there’s always an exception to the rule, as GC had traded slightly negative right after QE2 was announced by the Fed in November 2010. It’s my belief that was more a function of traders worried that GC would disappear from the market, without an actual fundamental shortage.
* More on the 3.625% 5/13 next week at the “Repo Roundup”