My core thesis right now is the US G-SIBs are being stuffed to the gills with coupon USTs. 1/x

I've done a rather poor job of describing the situation that gave rise to said stuffing, and will try to rectify that. Specifically, I've conflated description of primary dealer and bank portfolio gill stuffing. 2/x

When you aggregate all UST holdings from US G-SIBs, you get the following (note that only $JPM has 3Q filings are complete, but others should w/in two weeks). As you can see, $JPM has added $152bln of USTs in last 4Q's, but they're not alone. 3/x

Every G-SIB save for $BK and $STT (which are custody banks, so special cases) have added double-digit billions of USTs in prior 3Q's, likely more when the new data is available. 4/x

The data above are BHC aggregates. Here's what I've done a bad job describing and defending: A decent amount of this stuffing comes from PDs' trading accts, but another big piece is HQLA portfolios. Note 3Q 2019 data is $JPM only. 5/x

In fact, based on @federalreserve holdings data (more frequent than the above qrtly fin'l filings), what is looks like is that PDs have been selling trading assets into banks' HQLA portfolios. Weird huh? We'll have more evidence when the other G-SIBs file. 6/x

"Why" PDs and bank HQLA portfolios are loading up with USTs is complicated, but essentially my view is that they're the cash asset buyers of last resort. And assuming these entities follow their usual gameplan, they've hedged most interest rate risk. 7/x

Just b/c they're essentially forced buyers doesn't mean that PDs and banks' aren't profiting off of UST holdings. Negative swap spreads mean 5yr UST can be swapped back to L+3.5bps. 8/x

While I doubt many G-SIBs want this big of PD/HQLA UST holdings, it's tolerable, as the L+3.5ish spread can be levered up many times, especially since foreign funding of the repo market is quite willing (even if foreign buys of coupon USTs are low). Yup, I said repo. 9/x

In any case, BHC aggregate coupon UST buys have their limits, whether they're done at the PD level (where obligated to buy) or at the HQLA level. At the current rate of net adds ($50bln/mth) funded by repo, those limits are defined mainly by LCR at the BHC level. 10/x

USTs are 100% HQLA for the numerator of the ratio but repo outflows is bad for the denominator of the ratio, and funding term UST with o/n repo decreases the LCR. So basically there's a limit to how many USTs US G-SIBs can absorb. 11/x

At the current rate of UST buying (~$50bln/mth), G-SIBs will find LCR limiting in 1Q 2020. Meanwhile @federalreserve has started buying up T-bills, seemingly missing out on the fact that intermediate to longer coupon USTs are the real problem. 12/x

Economically speaking, the risk is that BHC aggregate balance sheet holdings of USTs crowd out BHCs' ability to lend to private credit markets. That includes PDs providing liquidity to bond/loan transactions and banks providing net lending at the margin. 13/x

Why? The easiest way to improve LCR ratios in the face of growing coupon UST holdings is to eliminate non-HQLA assets which require funding, such as corporate bond or loan holdings. Recent big underperformance of lev loans and CCC credit may be a piec of that trend. 14/x

If that theme continues, @federalreserve views LCR tightness --> credit contraction not as a technical problem of excess UST supply on BHCs, but as evidence of deteriorating financial conditions...and we get more insurance based rate cuts as soon as 1Q 2020. 15/x

Funnily enough, these cuts also solve underlying prob of foreigners not funding $1.1tln budget deficit, as lower front end & steeper curve reduces FX-hedging costs AND incentivizes real money buyers to leave o/n repo mkts and come back into coupon USTs. 16/x

The path to that point might be volatile for risk assets, but there's no reason the Fed can't save the day again. Fin.

PS: I hope these comments clarify recent threads in which I've been unclear. CC:
@Barton_options
@kimgits1
@deepwatrcreatur
@dandolfa
@kimdriver11
@mattwaters
@an10nev
@MonetaryWonk