CARRY COLLATERAL SWAP IN J-BILLS
EDU DDA Oct. 25, 2024
Summary: Repo fails are up again in the latest week, continuing their rising trend which dates back to early August. The events back then were closely tied to Japan’s carry trade. We have even more powerful evidence pointing to this from the Japanese perspective: three-month J-bills. Those rates are behaving in a way we’ve seen numerous times before and, more importantly, like fails they’re still doing it here toward the end of October.
The Federal Reserve reported another small spike in repo fails in the latest weekly figures. It wasn’t huge, though at just less than $200 billion the total adds one more week to the trend reversal for fails dating back to early August. There is no easy explanation for where this collateral difficulty is coming from, though the timing (as I discussed on my last YT livestream) strongly implicates Japan.
One further reason to suspect this is the performance of Japan’s 3-month Treasury bill (J-bill) in behaving in an all-too familiar way. Going back more than a decade, this tenor more than any other has corresponded with broad eurodollar difficulties particularly those dealing with collateral. There were several very outbreaks during the rocky year of 2022, for example, and it continued in the early months of 2023 right as SVB and, more relevant, Credit Suisse found funding had become especially troublesome.
Its most explicit relationship with eurodollar illiquidity is easily early 2016. To begin with, the 3-month J-bill had been exhibiting signs from as far back as the first few months of 2013. You can clearly see the abrupt dip in the rate relative to the Bank of Japan’s overnight call rate (policy). This was even before QQE had been announced let alone begun, so it wasn’t an influx of yen bank reserves created by the central bank program.
As that showed up, there was an unfolding mess US$ repo. The Fed had unwisely been buying on-the-run USTs as part of its QE4 introduced a few months earlier at the end of 2012 (QE3 was started a few months before and had specifically targeted MBS, which is why I classify QE3 and 4 separately). There were a number of USTs trading special, a situation that occurs whenever demand for a specific instrument leads to repo rates lent against it tumbling below the rest of the market (cash lenders accept a lower interest rate when lending that cash to collateral owners who have the bond or note the cash lenders are in need of).
It was a surefire sign of collateral shortage and one I know personally (through members of Congress and even a few meetings their staffers had with Bernanke himself) had been related to the Fed’s QE buying (Bernanke shortly thereafter told them the Fed had stopped buying on-the-run). The knock-on from a collateral shortage would be reduced circulation therefore some desperation among other US$ borrowers out in the eurodollar world.
With Japanese firms, this kind of funding shortage leads them to borrow in FX markets which ends up creating more demand for Japanese bills and bonds as a byproduct. If repo funding in US$s gets to be expensive or collateral flow is difficult, then Japan’s banks and big financials can alternately borrow in FX swaps which means borrowing US$s collateralized by yen (the real carry trade). The US$ provider ends up holding yen which gets parked in J-bills or perhaps JGBs down the curve, just some place safe and liquid.
The eurodollar lender doesn’t necessarily care what the return is on the J-bill because they are compensated by the rate on the FX swap. In fact, the more desperate the Japanese borrower, the greater the premium (negative basis) the swap meaning it becomes highly profitable to lend dollars even if that means taking the yen collateral and investing in negative-yielding bills or bonds (same thing happened in Europe with negative-yielding Germans, for example).
So, trouble in US$ collateral from repo owing to collateral difficulties leads to increased demand for borrowing dollars in FX from Japan. For Japanese (and others such as in Europe), this is basically a collateral swap where the UST shortage leads Japanese to reroute into FX markets which allows them to substitute J-bills or similar to acquire US$ funding.
We see this link repeatedly throughout starting with those collateral difficulties which showed up in early 2013. It provoked a noticeable drop in the 3-month J-bill rate, not huge by any means yet clearly visible.
There would be another one six months later in September concurrent to the emerging emerging-market “currency crisis”, the first rumblings of what became Euro$ #3. By then, though, QQE was already well underway which meant systemically there was a steady and growing internal (to Japan) demand for bills that depressed yields on them if in more constant fashion.
Its lower trend in bill rates was from QQE, but the short run drops are US$ related, breaking the trend. That was the case leading up to October 2014, pushing the bill rate negative for the first time.
If you don’t remember October 2014, it was an especially important month for Euro$ #3 escalating primarily on collateral problems. On the 15th of that month, the US Treasury market would experience a “buying panic” which Treasury and the rest of its useless regulatory cousins (Fed and OCC) pinned on computer trading. It was instead one of the clearest, more straightforward examples of an acute shortage and outright panicky scramble for collateral.
Early in that morning, UST yields plunged with buying coming in from everywhere consistent with the collateral scrambles we’ve documented over the years. For the Japanese, their 3-month bill would drop under zero on September 12 as spreads widened in US$ junk collateral (sound familiar?) The rate would be back and forth, in and out of zero a couple times (including October 15 itself) culminating in a -0.045% rate on October 23.
There would be another short outbreak (seen from the 3m J-bill) in the middle of January 2015 right when these dollar issues forced the Swiss National Bank of break its peg with the euro (SNB directly complaining about the dollar when doing so).
More of these would happen later in 2015, especially November, but the big ones were saved from early 2016. A sharp drop in 3m J-bill rates coincided with the BoJ trying out NIRP, or negative policy rates, for the first time. But it was the second half of March and then April which sealed it.
To begin with, the FX basis had become hugely negative by then. That was an acknowledgement of what I described above, demand for FX as an alternative to more difficult funding in US$ repo as UST collateral all but stopped flowing. With regard that, repo fails absolutely spiked the week of March 9, 2016, soaring to $889 billion, highest since the worst days of 2008. Then they did it again the following week, March 16, rising to $873 billion.
With those difficulties in US$ repo, the FX yen basis had grown massively negative. I wrote the following on March 18 while (because) all this was ongoing:
The negative yen basis swap acts like leverage where even yields on the interim "investment" are negative. Any speculator or bank with spare "dollars" could lend them in a yen basis swap meaning an exchange into yen. Because you end up with yen you are forced into some really bad investment choices such as slightly negative 5-year government bonds, but that is just part of the cost of keeping risk on your yen side low. Instead, the real money is made in the basis swap itself since it now trades so highly negative. The very fact of that basis swap spread means a huge premium on spare dollars; which is another way of saying there is a "dollar" shortage. Because of the shortage and its premium, you can swap into yen and invest in negative yielding JGB's in size and still make out handsomely. There has been, in fact, a rush of foreign "money" into Japan to take advantage of this dollar shortage; the fact that there has been such enthusiasm and it still has not alleviated the imbalance proves scale and intractability.
I used the example of a 5-year JGB, though by early April it was registering with the 3m J-bill, as you can see on the chart. The week of April 11 the bill rate plunged all the way to -0.473%!
Whether or not that major funding squeeze was related to China and its attempt to stabilize CNY using proxy means is another story and still an open question. One thing that did happen which raised my suspicions is the Chinese suddenly stopped using their FX reserves (selling USTs) during both of these months, March and April 2016. I only raise this factor just as a reminder China is often in the middle of all these eurodollar woes, especially 2013 and forward (there is some evidence what provoked the collateral squeeze in February/March 2013 was China-related).
It took months for things to eventually calm down, not until early 2017 with the rise of “globally synchronized growth”, really Reflation #3.
That’s not to say everything went back to fine and normal. Japan’s 3m bill rate not to mention its FX basis would stay negative throughout, though no noticeable dives in bills. The next time we saw one was the first week in October 2018 – yep, the very start of the landmine. While China was closed for its Golden Week, Japan’s bill rate dropped from -0.127% all the way to -0.322%.
We’d see the 3m J-bill signaling more collateral and US$ funding problems August and September 2019 – yep, September 2019’s repo rumble. Its rate would actually turn positive briefly in early August and then start falling again right when the “recession scare” in the US (it was more than a scare in Japan, the global downturn would lead to a declared outright recession there) was inverting the 2s10s in USTs and causing Jay Powell to begin cutting the fed funds rates.
There would even be another plunge in the 3m J-bill to end that eventful September.
Having established the connections through repeated correlations, including, as noted at the top, all throughout 2022 and early 2023, I’ve noted the reversed trend in repo fails from early August when the yen carry trade reversed. At the same time, the 3m J-bill rate has been behaving in the same way as those earlier instances.
It had previously turned positive around March 2024 right when the Bank of Japan began to raise its O/N call rate, so chalk that up to the policy change. The 3m yield would remain on the upside of the benchmark right up until the last day of July and the first couple days of August when all the trouble went down – and the shift in US$ UST repo fails commenced.
The Monday morning of August 5, the worst of the meltdown in Japan and elsewhere, the 3m rate dropped to 0.075% when BoJ’s O/N rate was 0.25%. It would remain well underneath the BoJ all throughout August, getting as low as +0.055% on August 26. But then on September 16 it would fall even more, very nearly turning negative again by the start of October!
Remember, those were the very same weeks when repo fails surged, the weeks of September 25 and October 2. More to the point, as of today the 3m J-bill still yields almost zero while FRBNY just reported another weekly rise in fails (for one week further back).
We’ve had some of the most benign collateral conditions from the middle of last year right up until the end of July. Fails were falling as a trend, T-bills were abundant and rates in both Treasury and Japan bills were well-behaved.
That all seems to have changed in early August, including an obvious scramble for collateral on August 5. It doesn’t appear to have been a one-off. While collateral (and dollar shortage) indications aren’t going crazy, they are now to where they hadn’t been before. Plus, they are sticking around despite everything supposedly going right lately (at least according to some macro data).
To review: dollar funding problems including collateral difficulties cause Japanese eurodollar borrowers to swap collateral using yen rather than USTs by rerouting from repo into FX. The more the Japanese do this, the more yen gets used as collateral the more “extra” demand there ends up being for 3m bills (and other JGBs) even if they yield nothing or less than nothing.
It’s now the end of October and we continue to get rising repo fails plus at the same time the 3m J-bill yield is flashing the same related warning. Japan is involved (carry trade). The collateral problems are likely related to CLOs though there isn’t nearly enough information to make that determination.
CLOs would mean US recession risk. It’s fair to conclude that collateral conditions are no longer benign, therefore something to keep a close eye one.
