The Repo market is an interesting market from an economics perspective. Unlike traditional markets where there’s a supply and demand curve for a specific instrument, in the Repo market there’s two supply and two demand curves. Wait, don’t leave! I promise this won’t be an economics lecture. In the Repo market, there’s supply and demand for securities and supply and demand for collateral, all in the same market.
One way of understanding movements in Repo rates is by watching factors that influence cash and collateral. Of course, on the cash side, we’re not talking about all the cash in the financial markets, just the amount of cash allocated to the Repo market. Here are some major factors which affect the supply and demand for cash and securities in the Repo market:
Net New Treasury Issuance
It doesn’t have to be all about U.S. Treasurys, but that’s the main market and I’ll be concentrating on it. When the U.S. government budget deficit is growing, there are more Treasurys being sold into the market. That is, more new securities are being issued than maturing securities. That puts upward pressure on GC rates – the market is absorbing new supply on a regular basis. Conversely, when the Treasury was paying down debt, like in the early 2000s, there are net securities leaving the market on a regular basis. Back then, I remember some commentators discussing how the U.S. Treasury Repo market was going to disappear once the budget surpluses paid off all the outstanding Treasury debt.
I always look at the net new Treasury issuance numbers published by Wrightson each week. Large new Treasury settlements will usually put upward pressure on funding that day.
QE Programs And Other Buy-Backs
When the Fed is running a QE program, they’re buying Treasurys on a regular basis and effectively taking them out of the market. When the Fed is in the midst of QE buying, GC Repo rates slowly move lower from month to month. After QE2 ended in June 2011, GC slowly trended higher, hitting a high in September 2012 – ironically, right when the Fed started QE3. The graph below illustrates the progression from GC trading from below fed funds to trading well above it.
Some customers regularly pull their securities or cash out of the market on quarter-end, year-end, and sometimes even on month-end. Institutions are “window dressing,” making themselves appear to be less leveraged on their financial statements. It means there’s more cash chasing fewer securities, or vice versa. Keep in mind, there are changes in both cash and collateral on statement periods. Repo rates move that day in a battle between the changes in cash and collateral.
Make Up of Treasury Holders
Who’s holding what makes a difference. If the Street is long and has large inventory positions, those securities are in the Repo market each day. If hedge funds own Treasurys, it’s the same thing – those securities are floating around in the Repo market. One way to follow this is on the Fed’s website – Primary Dealer net financing positions.
Conversely, when “retail”* accounts hold Treasurys, often those securities don’t come back into the Repo market. One gauge of this number is the foreign holdings of Treasurys. When Treasurys trade special, there’s often market rumors about them being held by a Japanese insurance company – the standard default reason for why an issue is gone from the market.
A good example of “retail” ownership occurred with the 3.625% 5/13. The bond market was expecting an ease in June 2003 which never came. “Retail” buyers were long a ton of the current 10 year note and when the ease didn’t arrive, rates backed-up. Most of the 3.625% 5/13 – the current 10 year note – was trapped in hold-to-maturity portfolios which couldn’t sell them or loan them into the Repo market. The result was the longest natural squeeze in Repo market history. The moral of the story is: the more “retail” the ownership of Treasury securities, the fewer of those the securities come back to the Repo market.
The securities lending groups are the gatekeepers for the “retail” portfolios who loan securities into the Repo market. The amount of securities they have available to loan represents investor confidence in the overall market. When the market is efficient and there’s confidence it’s functioning smoothly, more end-user portfolios will loan their securities through the securities lending groups. Securities lending supply of loanable assets peaked with about $15 trillion available and about $3 trillion out on loan before the financial crisis. Those numbers dropped to about $8 trillion available and $2 trillion out on loan during the financial crisis. Five years later, we’re back to about $14 trillion available again.
The Federal Reserve runs a program allowing Primary Dealers to separate the principal portion of a bond from the coupon payments, it’s called “stripping.” The more Treasurys that are stripped, the less supply of regular general collateral there is in the market. Naturally, if the market gets distorted and too many bonds are stripped, the principal and coupon components can be “reconstituted” and put back together at the Fed. The amount of Treasury STRIPS is mostly a factor for individual Treasury issues, but it’s still a minor factor in the overall Repo market.
No one likes not receiving their securities; except, of course, Repo desks which are making money on “positive fails.” For everyone else, they worry when they’re getting failed to. On top of that, large scale fails often occur during major market dislocations which makes those participants even more worried at just the wrong time. When fails increase, investor portfolios stop lending their securities into the Repo market.
A flight-to-quality is a big driver of Repo rates, but only during certain times. During a crisis, “retail” customer buying will take Treasurys out of the market to be tucked away safe in their portfolios, not seeing the light of day until the crisis is over. In general, these customers do not loan those securities back into the Repo market. They’re worried about counterparty risk and just want their money parked in safe, short-term Treasurys until the problems blow over. The graph below illustrates GC Repo rates during the Liquidity Crisis of August 2007.
Short-Base In The Market
The amount of shorts in the market affects general collateral rates. The more the shorts, the fewer securities there are for general cash investors. Times when the short-base of the market is larger, it puts pressure on GC rates to trade lower.
Investor Appetite For “Cash”
When investors decide to stay in “cash,” they’re not actually leaving cash in their account, they’re putting it in the Repo market. There have been several times in past years when the market is waiting on a Fed move. Back in May 2000, no one was sure if the Fed would tighten at the May FOMC meeting. The result was a lot of cash parked in the Repo market, waiting to be allocated based on what the Fed did. The seasonal April collateral shortage was supposed to end by May, but instead it continued well into June that year.
When GC rates are close to zero, there’s less “opportunity cost” for cash to sit idle and not earn a rate of return. We saw this during the Debt Ceiling Crisis of July/August 2011 when investors were nervous about a technical default and decided to park actual cash in their clearing account instead of the 0% to .10% rates they could earn in the Repo market. Had rates been in the 5.00% to 6.00% range, those decisions would have been different.
Also, when GC rates get relatively low compared to other overnight investments, cash will move into the instruments with higher rates, like agencys or CM bills, and vice versa. It’s the natural market mechanism. For example, when GC was trading above the IOER rate (.25%) in September 2012, bank cash started moving into the Repo market. GC rates got “sticky” above .25%.
* By “retail” it’s not “mom and pop” accounts, I mean “buy and hold,” unsophisticated institutional accounts.