Quantitative Easing And The Repo Market
Yes, I know, I’ve been talking about the unintended consequences of the Fed’s QE programs on the Repo market for months. Warning of dire consequences if the QE purchases continue and overnight rates drop down to zero percent. Well . . . suppose I backup my warnings with some statistics and graphs.
QE1
The Fed first began buying Agency Mortgage-Backed Securities (MBS) at the beginning of November 2008, pretty close to the beginning of the financial crisis. The fed funds rate was not quite at 0.0%, but the Agency MBS market had broken down due to a combination of Fannie Mae and Freddie Mac collapsing and investors fleeing the mortgage market. The Fed purchases continued for 17 months and the Fed, all in all, bought $1.7 trillion in Agency MBS and Treasurys. By June 2010, the Fed stopped buying, the Fed’s balance sheet stood at $2.1 trillion, and QE1 was over. However, it’s impossible to measure the effects of QE1 on the Repo market, given there were so many other factors affecting Repo rates during that time.
QE2
The first hints of QE2 came from the Jackson Hole conference in August 2011 where it became clear the Fed was worried about deflation. QE2, which ran from November 2010 through June 30, 2011, had the Fed purchasing $600 billion in U.S. Treasurys and reinvesting the maturing principal and interest from the QE1 portfolio. Unlike QE1, QE3 and QE4, the QE2 program was specific, there was a specified amount and time frame.
During the QE2 program, the Fed averaged about $100 billion a month in purchases. The graph [above] shows the spread between General Collateral (GC) and fed funds. In October 2010, GC was averaging at .22%, and when the program was over, GC was down to .03 % in July 2011. In relation to fed funds, GC started off trading about flat (+.8 bps) to fed funds, and was averaging 6.6 basis points below fed funds when the program ended. The takeaway, at least from a Repo market perspective, is that a $100 billion a month QE program moved GC down roughly 2.7 basis points each month.
Keep in mind, there’s another factor involved – the amount of Treasury issuance. During 2011, the government budget deficit was large and the Treasury was issuing a load of debt. When the Treasury is increasing the amount of U.S. Treasury debt outstanding, it’s adding Treasury collateral into the market each month. Of course, during a QE program, the Fed is taking collateral out of the market each month, somewhat offsetting the increased Treasury issuance.
QE3
Two years after the QE2 program ended, the Fed was again worried about the recovery and wanted additional stimulus. On September 13, 2012, the FOMC announced a new buying program, this time limited to Agency MBS that would run at $40 billion a month. It would come to be known as QE3 – the third round of bond buying.
With little noise in the overnight funding markets, MBS Repo rates are pretty easy to isolate. It’s pretty easy to see the change in the spread between Agency MBS and fed funds in the Repo market from October 2012 to the present. When QE3 was first announce in September 2012, Agency MBS collateral was trading at 14 basis points above fed funds. Over the past month, Agency MBS collateral is now trading 3 basis points below fed funds. Note, the graph [above] shows four spikes in the spread: one for a hurricane and three for quarter-ends.
QE4
QE4 is the younger sister program of QE3, having been born just three months later on December 12, 2012. It encompasses $45 billion U.S. Treasury purchases each month, and all total, the combined programs account for a total of $85 billion fewer Agency mortgages and Treasurys in the market at the end of each month.*
And here’s what we’ve all been waiting for – the impact of QE4 on the Repo market. Obviously, as we know from QE2, fewer Treasurys are going to drive GC rates down, close to 0.0%, and push the GC/fed funds spread into negative territory. Before the QE4 program started in December 2012, GC was trading at .24%, which was 7 basis points above fed funds. In the past month, GC is down to averaging at 5.5 basis points, which is 4.3 basis points below fed funds. It took approximately 7 months of Fed QE purchases to move GC rates from .24% in December 2012, down to 5.5 basis points. Surprisingly, and still a rough calculation, GC moved down an average of 2.6 basis points a month over the past 7 months. That’s a pretty similar move to QE2, and I promise, I didn’t massage the numbers.
The Problem
We now have a rough estimate that QE2 moved GC rates down 2.7 basis points each month and another rough estimate that QE4 moved GC rates down 2.6 basis points. The problem confronting the Repo market is that overnight rates are averaging 5.5 basis points right now. With continued QE purchases, there is no doubt that General Collateral will be at dead zero by November. Once GC rates hit 0%, they have no room to go lower. Why do zero percent overnight rates matter? When there’s no interest rate on cash, it will create distortions in the Repo and funding markets, let alone what will happen when there’s continued pressure for overnight rates to go negative, but they can’t.
Here’s an example. Suppose you’re a cash investor and the market is paying 0% on overnight cash. Why would a money fund ever do a trade and receive no interest? They have to write a ticket, take counterparty credit risk, process trades, and possibly pay regulatory capital charges. They’re better off leaving their cash uninvested at their clearing bank.
Now, not investing cash in the Repo market is certainly a problem, but the clearing bank also has a problem. They have to pay the FDIC insurance tax on their customer’s cash balances (deposits) and that costs them 12 or 13 basis points. The last time we were here, Bank of New York Mellon began charging customers on cash balances above $50 billion in August 2011 – right after the last QE2 program pushed rates down to 0% in July 2011. That’s a great example of one market distortion, however, as there’s pressure on rates to go below zero, I am confident other distortions will emerge.
Reaction By The Fed
On May 22, Ben Bernanke first mentioned the idea of “tapering” – slowing down QE3 and QE4 bond purchases. Now, I’m pretty confident he was not addressing my dire warnings of damage to the Repo market, but I believe the Fed is concerned. Recently, the New York Fed announced they had tested Reverse Repo operations and stood ready to use them to drain liquidity “as the FOMC saw fit.” In addition, they stressed that “These operations do not represent a change in the stance of monetary policy.” That’s good news for the Repo market!
Specials
The new 10 year note (2.50% 8/23) settled just last week and I already know what’s on everyone’s mind – whether it will trade as special as the February and May 10 year notes. When you have the short-base of a triple issue moving into a single issue, watch out!
The new 10 year note will, most probably, follow the same trading pattern as the past two 10 year notes. It all goes bak to the auction cycle: 10 year notes are on a three month auction cycle with reopenings during the two intervening months. That means the new 10 year note, which just settled on August 15, will be reopened on September 15, about doubling its size outstanding. The final reopening will settle on October 15, then making the 2.50% 8/23 a “triple issue.” As a rule of thumb, 10 year notes trade the most special during the last two weeks of the first month after it’s issued. That makes sense, of course, based on pure supply and demand – there’s simply not much supply outstanding, and if short-demand is significant, a 10 year note will trade very special, just like the past two 10 year notes.
Given current Repo market conditions, I don’t expect the 2.50% 8/23 to trade as special as the past two 10 year notes, but I do expect it to follow the same trading pattern.
*Also note, as the Treasury began issuing fewer Treasurys beginning in April/May, decrease in new supply of Treasury collateral, it makes the effects of QE4 purchases all the more poignant.
