Projecting the X-Date and the TGA
TLDR
X-Date very unlikely in June
I’ve built a model that projects every component Treasury withdrawal and deposit for each day in May and June and included that output in this post
TBAC refunding deficit projections imply an earlier July X-Date and the slight possibility of an early June X-Date
My modeled closest approach in June is June 9th with TGA at 39b
*** Post Edit 5/19 8:00 AM
I missed something material in my original analysis (below this edit). I did not account for the discount on newly issued bills and to a lesser extent the interest increment on United States savings securities and the increment on inflation-protected securities (all found in table IIIB of the Daily Treasury Statement). My rough projection on these is an impact of -21b through 6/1, -23b through 6/6, and -25b through 6/8, so now calling for closing TGA balance of 42b on 6/1, 23b on 6/2, 28b on 6/5, 21b on 6/6, 23b on 6/7, 15b on 6/8, and 14b on 6/9. My core thesis of no x-date in June is intact but second week of June will be perilous indeed. There is little to no room for my TGA withdrawal/deposit model to be wrong. Apologies for the miss, Ill detail more in a post later this week, but wanted any new readers of this post to be aware.
***
Every year on baseball’s opening day, my best friend and I will exchange a series of texts that go a little something like “Holy cow dude, did you see Aaron Judge hit a home run today. He’s on pace for 162 homers this year!!!!!” replied to with something like “That’s crazy, and even more crazy is Mike Trout with 4 RBIs, he’s on pace for 648 RBIs this year!”. There is just something about abusing the pace of something over a short time period to project absurd long term outcomes that we find hilarious.
So I smile a bit every time I read a breathless article about the pace of TGA decline due to a large drop the prior day projecting it could be exhausted “by next week”. Of course, in the macro sense they aren’t wrong; the TGA is certainly dropping in the aggregate with spending outflows greater than taxing inflows and an inability (aside from extraordinary measures) to issue debt to make up for it. It will eventually run out. But like baseball statistics, the TGA’s withdrawal and deposit outflows are not uniform over time, some days withdrawals massively exceed deposits (as commonly happens on the 1st of the month), other days it’s the opposite (mid-April for example), and for some days the withdrawals and deposits are roughly balanced.
How does one project this X-Date?
Thus, to project when the debt ceiling X-Date will occur you have to project as accurately as you can all the component deposit and withdrawal flows of the TGA going forward (many of which march to the beat of their own transaction pattern drums) and then apply that against a starting point with a known balance of funds remaining (TGA level on that date plus remaining borrowing/extraordinary measures). To solve the first part, I have built a model that projects every TGA component withdrawal and deposit flow for each day from May 1st (I project 5/1 through 5/11 so folks can compare projected vs. realized deposit/withdrawal flow numbers for those days) through the end of June. I am publishing those daily projections in the attached spreadsheet. To establish the baseline, we will use May 10th since the Treasury conveniently detailed Friday that there were 88b in extraordinary measures remaining on that day. With a TGA closing balance of ~155b on April 10th that means the US govt effectively has 243b from then to tide it through the aggregate deposit/withdrawal deficit until those flows turn aggregate positive for a while in mid-June (mini tax receipt period) and ultimately to the ~143b additional extraordinary measure that unlocks on June 30.
Is the X-Date in June?
Will the TGA make it to the end of June before exhausting (and thus to some point in late July or August)?
Yes! I think so, the output of my model tells me it will anyways, but it will be waaaay too close for comfort, and there is significant disagreement between what my model projects for remaining deficit between May 11 and June 30th (229b) and what Treasury suggests is remaining (288b) based on the early May updated refunding statement. More on that later…
How things play out in June
First though, here is how I think it will play out using my model output:
· Treasury will use all 88b of the available extraordinary measures between now and June 1st (mixture of net bill, including cash management bills, prior to and coupon issuance on June 1), so after that the deficit will have to fund entirely from TGA balance
· June 1st is a massive deficit outflow (withdrawals vs deposits) to the tune of ~73b (OMG! that projects to like 1.5T of deficit over the 21 days the TGA is open in June!!!!!! Joking, of course, couldn’t resist) which will be reduced by whatever amount of the 88b in EM is left to deploy in a large debt issuance the same day, ultimately ending with a TGA balance ~63b on June 1st, the start of the final leg of the descent to the closest approach to exhaustion in June
· Friday June 2nd is also a large deficit outflow day (a large batch of social security checks go out that day) to the tune of ~19b. So TGA further drops to ~44b
· TGA stays perilously low the week of June 5th drifting a little lower
o 5th: +7b TGA at 51b
o 6th : -7b TGA at 44b
o 7th : +2b TGA at 46b
o 8th : -6b TGA at 40b
o 9th : -1b TGA at 39b
· Mid June tax receipts start rolling in week of June 12th up 94b over the week with the biggest inflow on 6/15 but inflows or flat every other day. TGA ends the week at 133b
· Week of June 19th is net outflows but mild, ~13b over the week so TGA drops back to 120b end of the week
· Last week of June starts with ~24b of aggregate outflows, so TGA at 96b on June 29th
· June 30th is another massive outflow (82b)day (many of the large first of the month payments for July go out on Friday June 30th since the 1st is on a weekend). So uh oh, does that mean the TGA is going to 14b end of June 30th and truly enter the danger zone…….pausing for dramatic effect…….
· No! Fortuitously, ~143b in additional extraordinary measures unlock on June 30th. With the net coupon issuance scheduled for that day they should use ~31b that day so maybe a TGA at 45b end of June 31, but Id expect Treasury to use the rest of the 102b in unlocked extraordinary measures/bonus borrowing early in July
· So July starts with ~157b to fund its deficit (albeit with the big first of the month payments out of the way).
TBACs numbers imply an earlier date
As I stated earlier though, there is the pesky issue of a 59b difference between what my model projects is the remaining deficit through the end of the qtr and what the TBACs latest refunding statement implied it would be.
Let’s review how we get to 288b of remaining deficit (for Treasury). From the TBAC,
178b (TGA Start of 2nd qtr Balance)
+ 726b (Net privately held marketable debt to be (though it actually wont) issued in the qtr)
- 550b (TGA End of 2nd qtr Balance) (also wont actually reach)
- 180b (UST QT between March 1st and June 30th)
= 174b of budget deficit for the 2nd Qtr
Through end of May 10th, the non-debt deposits/withdrawals have actually been a ~114b surplus so add that to 174b projected deficit for qtr and Treasury implies we have ~288b to go.
How reliable is that number? Good question. On the one hand you would think it should be good. Surely, Treasury has sophisticated models that should mostly get it right (surely way more sophisticated than what Ive put together). On the other hand, between February 1st and May 1st, those same sophisticated models had to adjust their deficit projections for this quarter up by 117b. That’s a big adjustment meaning either a big miss on their revenue projections (and thus their models) or perhaps a little bit of politics in play as well to paint as worst case scenario picture as possible to motivate debt ceiling talks. I dunno.
X-Date Still unlikely in June
So if Treasury’s deficit projections are right does that mean we breach on June 2nd ? No, not necessarily, it would depend on where my model of the next two months underestimates the deficit. If it underestimates the deficit (higher component withdrawals/lower deposit withdrawals) entirely in whats left of May or early June then yeah, a breach/X-date between June 2nd and 9th is possible. But, to the extent my model has underestimated the deficit to the end of the qtr, I think its much more likely that it is overestimating the tax receipts in mid June. I modeled these in large part using last year’s numbers. Perhaps it will prove too generous. If Treasury is correct and my error is largely there than we still comfortably make it to the 143b lifeline on June 30th. Of course, then July starts with closer to ~100b.
To the extent my model is correct on the deficit though, it cant be in June. So when specifically will that date be? I will make a specific projection a little later for two reasons. 1. My model currently only runs through end of June, most of my projection methodologies for the individual deposit and withdrawal categories are generalized based on past data and transaction patterns, but I hard coded in a few projections that don’t work like that (e.g. calculating interest payable on treasury securities) which will take me a little time to modify, and the second and larger reason is because Id like more real data under the models belt both to have the real starting point (where the TGA actually is in the first week of June) and because my model uses year over year increases/decreases in transaction levels using the prior month to the start date of the model run on a number of withdrawal/deposit categories. So final May data will make the July (and August) withdrawal/deposit flow projections better. Ill forecast a specific x-date in early June when I have that data and my models resulting output and update after the June tax receipt data is in. But in a general sense. The first of the month deficit outflows on Aug 1 and the lesser but still large deficit likely on Aug 3 due to the first big social security payment batch for August look like big hurdles to overcome.
The model I have created to project treasury withdrawal/deposit flows
My original intent with this post was to go into depth on the model I have built to project the underlying deposit and withdrawal flows for the TGA. I will eventually do that (perhaps later this week, perhaps next). For now, I have posted the models projected daily deposit and withdrawal flows for the TGA for every day from May 1st through June 30th. Every assumption in my analysis outside of the amount of extraordinary measures remaining is baked into the model output. Through May 11 the model is running a little behind actual on both withdrawals (7b behind: 311.6b projected vs. 318.4 actual) and deposits (9b behind: 157.9b projected vs. 167.2 actual). There is still work and tweaking to be done to it I am sure and I will continue to do so, but I think its good enough now to support this analysis and thus I use it.
I will post regular updates on twitter in the coming weeks comparing my projected output vs. realized TGA numbers and their implications for the x-date. I invite everyone to follow along and welcome feedback and critique on my model’s output vs. reality. That will be easier to do once Ive gone more into depth on the various projection methodologies Ive used but if you look into the output and see something that looks off, Id love the feedback.
One last quick point on withdrawals to Federal Deposit Insurance Corp (FDIC) – I hard coded in the 12b in withdrawals earlier this month. I cant predict what these withdrawals or deposits back from might look like, so I predict no special withdrawals going forward and hard coded in the actuals that happened in early May. If there are future big withdrawals, my projections for withdrawals will be light by the same amount. It’s a wildcard in the analysis.
As always, thanks for reading!
John
Great breakdown. Thanks for all the work you do John.
Nice work John. Really helpful to have this as a baseline going forward.
hat tip-