The Fed has brought their 4-week moving average for Not QE back to $20 billion. Not all of Not QE is making it into reserves.
I believe this all ends when the Fed hits $500-$600 billion in Not QE. Beware the taper.
The Fed has become the lender of first resort in the overnight markets.
The term structure of the overnight markets is upside down.
This is all very bullish for equities until the taper.
How Far Can They Go?
The Fed kept the 4-week moving average of Not QE unchanged last week at exactly $20,050.25 million. When I told people this on Thursday, the day the weekly Fed balance sheet comes out, the first question I got from everyone was, “How far can they go?”
The only for-sure answer is, as far as they want to. But my best guess is that the number is in the range of $500-$600 billion. Not QE is currently at $346 billion through last Wednesday, which means another 8-13 weeks at current rates, or April 1 through May 6. If they stop in that range, it would be larger than QE 1, but smaller than either QE 2 or 3.
The Fed has $2,427 billion in Treasuries on their balance sheet, or 2 weeks short of eclipsing the previous peak before QT in October 2017.
What makes me believe this is that the year-spanning overnight interbank lending market, AKA New Year’s Day, was propped up by a total of $502 billion in Fed liquidity, roughly 50-50 Not QE and repo of varying lengths. This was sort of a stress test of the system. Not only is that one of the highest demand overnights for repo, but the Treasury had also stocked their checking account by $273 billion while this was all going on, which comes out of reserves via tax payments and bond sales. The Fed had “at least” $750 billion ready to go, but $502 billion did the trick.
This past week they were at $516 billion total Not QE plus repo, creeping up more slowly than it had been, but still creeping up. The Fed has cut back drastically on repo — down to $170 billion this week from the year-end peak of $256 billion — $85 billion less. But at the same time there was another $99 billion of Not QE. Netting it out, the Fed is slowing this down:
Expect that line to start popping up a bit to the ~$12 billion range if the Fed remains on their $20 billion per week pace of Not QE. But it will still stay below the net pace of the rest of this chart.
So I believe how this ends is that repo eventually goes to zero when Not QE reaches somewhere in the $500-$600 billion range, and then it all stops. If previous QE rounds are any indication, the Fed will ease out of it, tapering slowly so that there are no big surprises.
Quick Data Note
In previous repo articles, people have asked where all the data comes from. Mostly from the Fed.
- The quarterly data taking us through Q3 2019 that covers sectoral balance sheets come from the Fed’s Z.1 tables, specifically 108 which covers the banks, 210 which covers Treasury assets, and 211 which tracks MBSs. We will not see Q4 data until March 12.
- The weekly data from the Fed’s balance sheet comes from the Fed’s H.4.1 tables. This gives weekly Wednesday levels for, amongst other things, Not QE, repo, reserves, and the Treasury’s checking account. H.4.1 comes out Thursdays at 4:30 Eastern time.
One thing to note about tables 210 and 211 is that there is often a large “instrument discrepancy” between the securities that are out there, and the ownership accounting of them in different sectors. This indicates there is some error in these numbers.
Also, you may notice that the numbers in the “Where the Treasuries Are” section are slightly different from previous articles. I changed the start date from the end of Q1 2014 to Q2, since this was closer to the peak in reserves.
Back In Business
After a lull at the end of the year and the first few weeks of 2019, the Fed had brought their 4-week moving average for Not QE back to $20 billion the week ending January 29, and it stayed there this week. The Fed will only have to add $15 billion this week to stay on course.
Since they really got cooking in October, they remain at about 79% bills. This is what the cumulative progress looks like:
You can view this chart as the three major sources pulling on reserves. Not QE (blue sections) and repo (green sections) add to reserves; whereas the Treasury’s checking account (yellow sections) withdraws from reserves when they stock up. As you can see, repo has begun to shrink, but the top half of the chart has been above a half trillion 4 out of the last 6 weeks, and I don’t anticipate it dipping below again.
But not all this liquidity is making it into reserves.
The green line is the top half of the previous chart minus the bottom half - the net to reserves from policy. As of last Wednesday, only half of the provided liquidity that wasn’t sucked up by the Treasury account made it into the bank’s reserves. Where’s that other $138 billion gone?
Just this latest week, the Fed bought $18 billion in Treasuries from the banks. The Treasury reduced their account by $29 billion. That’s $47 billion that should have been added to reserves. Only $40 billion made it there. Where’s the other $7 billion, and the other $131 billion that came before it?
The Fed keeps telling us that they are in charge of reserves, not the banks. The banks are quick to agree. Ask yourself: if the Fed is in charge of that blue line, is that chart what the Fed wanted, and if so, why?
Not QE and QE 2
I believe that we are in a situation very much like QE 2. Here’s the same chart as the previous one, but for QE 2:
From July 2010 through the end of that year, the gap between what was being fed into reserves and what wound up there was $183 billion. In that same period, the banks added $99 billion in debt securities, $65 billion in direct foreign investment, and $24 billion in equities. There’s a lot more in there of course, including about $140 billion of toxic MBSs and derivatives that vanished in those two quarters. My favorite part is the Fed bought $245 billion in Treasuries from the banks, who turned around and bought $282 billion in Treasuries, for a net for $38 billion.
But that $24 billion in equities took their exposure from 0.36% to 0.54%. This is all very small on their balance sheets, and even smaller compared to US market cap, $15.6 trillion at the beginning of that period.
But then it was $19.2 trillion at the end of the six months, a jump of $3.5 trillion:
How does that happen? What I think happened here is not the nefarious scenario painted by some of the Fed directly feeding the stock market, but that the liquidity they provide primes the pump.
- The banks increased their exposure to all sorts of assets, mostly debt securities, and most of those are agency debt and MBSs. But they also increased their exposure to equities, in QE 2 by 50%.
- Still, this is very small, only $24 billion in a vast pool of equities. But it also signals other market participants to increase their risk exposure, so they follow the leader.
- It also gives the banks more liquidity to finance leveraged positions.
This is how The Everything Rally happens.
Every cycle is different, but at the end, they look the same in one respect: everyone chases that thing that made them money earlier in the cycle. In this cycle, Follow the Fed was the most lucrative trade in equities, and it is the most lucrative trade right now.
But if I am right, and the Fed will start tapering off near $500 billion in Not QE, we could see a large correction or even crash when that happens. In QE 2, the Fed began tapering in July 2011, basically putting a stop on it by the end of September 2011. This is what happened to the S&P 500:
Total market cap fell from $20.2 trillion in Q2 2011 to $17.1 trillion by the end of Q3. The banks? They reduced their exposure to equities by 29% that quarter.
The taper is coming. Be prepared for it.
QE 4 Is Almost Inevitable
The Fed is leaning heavily into the short-term nature of their Treasury purchases to emphasize that this is Not QE. The original QEs were to allow the federal debt to rise quickly in 2009-2011 without spiking rates across the curve. This has narrower purposes:
- Bring down the short end of the curve so it would un-invert. This worked, until it didn’t. As of Friday, the spread between the 10-year and 3-month is 4 bps, down 31 bps in the past month.
- Alleviate the stress on the overnight markets.
I find it hard to believe that the Fed will be able to allow these bills to expire, and in fact, they have already replaced many billions. They have $169 billion expiring in the next three months. If they let that debt back into private markets, the overnight markets will return to turmoil, and the short end will spike.
On top of that, the federal government is adding more debt to the pile every week, $1.2 trillion in the 2019 Q3 TTM. I just don’t see how they expect to unwind this, when the overnight banking system couldn’t handle QT, where the Fed only averaged selling about $5 billion a week. Letting the short-term bills expire would be at a rate of about $16 billion a week.
They will need to replace every dollar of expiring bill, as they have already been doing, and eventually it will make sense to add term. But why does the Fed have to keep the Treasuries on their balance sheet over $2.5 trillion?
The problem is that at current rates, foreign buyers and domestic non-financial sectors are just not interested in buying the supply, and the banks and the Fed wind up having to backstop it. The repo turmoil is a symptom of this.
Where The Treasuries Are
From the end of Q2 2014 through the end of Q3 2019, the federal government added about $4.5 trillion in new Treasury debt. In this period, there were huge changes in the composition of who owns that growing pie, at $18.6 trillion as of the end of September. (This is only Treasury debt; the full tally for the Feds is $22.7 trillion — the rest is mostly agency debt.)
Reducing Their Piece of the Pie
- Foreign Buyers: -7.2 pp
- The Fed: -6.2 pp
- State/local governments: -1.0 pp
- Insurance: -0.2 pp
- Total Negative: -14.5 pp
Domestic Non-financial Sectors Taking More
- Households: +5.5 pp
- Pensions: +1.1 pp
- Funds: +5.2 pp
- GSEs: +0.4 pp
- Total Domestic Non-financial: +12.2 pp
- Banks/Brokers/Dealers: +2.0 pp
- Credit unions, foreign banks in US, etc.: +0.3 pp
- Total Domestic Financial: +2.3 pp
Percentage of total Treasury debt, before and after:
Foreign and the Fed went from a combined 62% of Treasuries to 48%, and domestic non-financials greatly increased their share. But the light blue slice, domestic financial sectors, had the largest relative rise. That 2.3 pp may not seem like a lot, but that is $600 billion, pretty close to what I think that overnight liquidity hole is. Moreover, $249 billion of that came just in Q4 2018, which was the prelude to the first repo hiccup in the 2018-2019 overnight.
In that quarter, there was $424 billion in new Treasury debt, and the Fed sold another $57 billion in QT for a total of $481 billion in new Treasuries out there. Here’s who took it:
The residual not taken by foreign and non-financial sectors, $249 billion, got backstopped by the domestic financial sector. According to the Bank of International Settlements, that $249 billion in Treasuries went almost entirely to the largest four banks, presumably JPMorgan (JPM), Bank of America (BAC), Citi (C), and Well Fargo (WFC). They had been providing almost all the liquidity for the overnight system, and since Q4 2018, they have been very tight. Whenever they are tight, the whole system is, and the Fed needs to step in.
So that’s where we were at the end of September when this mess really got cooking. We won’t get Q4 numbers on all this until March 12, but the shifts in all this will be dramatic.
“Mortgage-backed security” is actually a pretty broad category. It got a bad name in the financial crisis and for good reasons, of course. But there are very safe ones and less safe ones. Generally speaking, the very safe ones are the ones bundled by GSEs (Fannie (OTC:FDDXD) and Freddie (OTCQB:FMCC)), and other government agencies.
There is a very large misunderstanding about the GSEs’ roles in the financial crisis. Up until 2006, the GSEs were very conservative in the mortgages they bought and bundled. These loans conformed to a certain set of standards, and are known as “conforming loans.” As we know now, the banks were not as circumspect.
Through that cycle, the GSEs lost market share because of this limitation, going from bundling 70% of mortgages to 35% in just a few years. They requested from Congress that they be allowed to jump into the toxic pool, and Congress allowed them in 2006. Their leadership acted like a private firm, not a GSE, in that they were profit-maximizing, and ignoring their equally important fiduciary responsibilities.
The pre-2006 MBSs from the GSEs performed very well in the crisis, but they had enough bad ones from 2006-2007 that they needed bailing out too. After the crisis, they went back to only buying conforming loans. So the MBSs we are discussing here are the very safe GSE ones that are composed only of conforming loans. The banks have relatively few of the bank-bundled MBSs left on their balance sheets after TARP.
I bring this all up because in the same period that the banks added that $600 billion in Treasuries, they also added roughly the same amount in MBSs. They have an advantage and a disadvantage over Treasuries:
- Pro: Higher risk-adjusted return than Treasuries of equivalent duration.
- Con: They only count for 85% towards calculating the banks’ Basel III reserves, unlike Treasuries that are counted dollar-for-dollar.
When cash reserves are high like they were in 2013-2016, that con doesn’t matter. The banks had plenty of cash to meet their Basel III requirements. But when they get tighter like they have been since Q4 2018, it very much does matter.
This is why there has been growing demand for MBS repo. All else being equal, it is better to use MBSs as collateral:
- If you put a billion dollars in Treasuries into repo, the net to your Basel III reserves is -$50 million. +$950 million cash (assuming 5% collateralization discount), and -$1 billion Treasuries.
- If you put a billion dollars in MBSs into repo, the net is +$100 million to the Basel III reserves. +$950 million cash, and -$850 million MBSs ($1 billion minus 15%).
But if you think of it from the lenders’ side, they lose $100 billion from their Basel III reserves. Again, when cash reserves are very high, this is not a problem, but when they are tight, it is. That’s why Treasuries tend to dominate in private markets.
But there is growing demand for MBS repo from the Fed, which does not have Basel III reserves, so they don’t care what collateral they take. That MBSs are preferred by collateral holders, and one lender doesn’t care either way, means there has been growing demand for MBS repo from the Fed.
Where the the blue line is going above the green line, banks are submitting more MBSs than the Fed is willing to accept. As you can see, in January, the Fed raised their MBS offerings considerably until they were able to get it more in line with demand towards the end of the month, and then let it come down. But that demand is spiking again, with two hugely oversubscribed 2-week term operations last week.
This is all important because of the next section.
The Fed Has Become the Lender of First Resort
When this all began, my quick policy fix was for the Fed to have a permanent backstop facility that charged an extra basis point or two over the private market low rate. This way, the private markets would be the first choice, and collateral holders would only go to the Fed if rates were spiking in the private market. Not only does this keep the Fed the lender of last resort, it would also properly set expectations not just for next week, but next year and beyond.
But they decided to go another way. Every few weeks, they tell us what they will do in the next few weeks, and hint at what comes next. The repo contracts are part of the same market as private lenders, with the same rates.
In the first place, this is an awful way to set expectations. But it has also led to the Fed becoming the lender of first resort in the markets, and the evidence is in the MBS repo market. Once the banks figured out that the Fed would accede to their demand for more MBS repo, and that unlike other lenders it didn’t matter to the Fed either way, MBS repo from the Fed looks like their first choice.
What happened in the two wildly oversubscribed term repo operations on February 4 and 6 amplifies this. In total, the Fed only had $6 billion combined on offer for MBSs those day, and banks submitted $39 billion in collateral. But the spread between the high and low rate was only 3 bps on those days, which tells me no one was that desperate or it would have gone much higher. They were just shopping for the best deal, saw it was a crowded trade and pulled back.
The Treasuries were also oversubscribed, $78 billion submitted for only $54 billion in repo. Here the high-low spread was only 1 bp. No one is desperate for anything, they are just going to the Fed first.
The whole system is being further undermined by the term structure, which is upside down right now.
The Fed pays interest on excess reserves (the IOER). This is low, but nonzero and around inflation. It is there to encourage more cash reserves.
Effective Fed Funds (the EFF) and the repo rate (the SOFR) should be higher than that, by just a few basis points. This gives banks with excess reserves the incentive to provide liquidity for the overnight markets.
But this is all upside down right now. The IOER is at 1.6%, but the EFF and SOFR are lower.
Those lines are the weekly average spreads between EFF or SOFR, and the IOER. Think of it as the incentive lenders have to enter the market. Lenders were getting 10 bps over IOER as late as last fall, which is a pretty big incentive.
Now those spreads are negative, meaning it makes more sense to just keep your cash reserves in your Fed account, and take their interest. Not only that, as of Friday, the entire Treasury yield curve through the 10-year was inverted with IOER, which should be the lowest rate of all, not the highest.
No one is buying Treasuries for the interest. They are all expecting this to keep coming down.
Upshot for Markets
As I said, it looks to be like we are in a situation like QE 2. With the banks leading the way, everybody is pouring into equities, and so long as the Fed keeps their weekly average at around $20 billion, that should keep the rally going, virus be damned.
But the taper is coming, probably early spring if I am right about the $500-$600 billion target for Not QE. Be prepared, because that first step will be a doozy.
Previous Repo Coverage
In reverse chronological order:
- Repo And The Man: A Half-Trillion Later
- Repo: The Abundant Reserves Regime Is Broken
- Another Repo Hiccup Highlights Year-End Uncertainty
- Repo And The Man: The Fed Is Back To Managing Liquidity
- While You Were Sleeping: Understanding Repo Rates And What Happened Last Week
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.