Among the jewels: Treasury Department paid 88 cents on the dollar to buy back 10-year Treasury notes sold at auction Aug. 2020 at a record-low yield of 0.677%.
By Wolf Richter for WOLF STREET.
The Treasury Department’s buybacks of Treasury securities — it kicked off that program in April — show on an anecdotal basis what is going on in the market, including the bloodletting among longer-dated Treasury securities due to rising yields since mid-2020.
So far in October, there have been four buyback auctions – one per week. The Treasury department announces beforehand what securities by CUSIP Number it offers to buy back. The Primary Dealers submit their bids at what prices they’re willing to sell. Those bids are evaluated, some are accepted, others are not.
In total over those four auctions in October, the Treasury bought back $7.0 billion at par value (face value) in securities. But that’s not the amount it paid. It bought back many issues at a substantial discount to par value. As we can see, the buyback amounts are relatively small in the huge universe of the Treasury debt outstanding.
| Treasury Buybacks in October, amounts at par value | ||||
| Auction Date | Issues eligible | Maturity date range |
Offered million $ |
Accepted million $ |
| 2-Oct | 26 | Oct 2025 – Jan 2032 | 1,384 | 235 |
| 10-Oct | 26 | Oct 2029 – Sep 2031 | 4,963 | 2,469 |
| 16-Oct | 49 | Oct 2027 – Sep 2029 | 10,257 | 4,000 |
| 23-Oct | 18 | Jul 2032 – Feb 2054 | 977 | 323 |
| Totals | 17,581 | 7,027 | ||
We’re going to look at a few specific bond issues that the Treasury department bought back at each of the past four auctions.
On October 23: bought back TIPS:
Holders of TIPS (Treasury Inflation Protected Securities) get paid interest every six months at a fixed rate. In addition, the inflation protection amount, based on CPI, is added to the principal of the TIPS. So with inflation, the par value of the security grows, and the percentages here are off par value that includes the inflation protection amount added to the principal. Here are four of the issues it bought back.
At 62.04% of par value, bought back $10 million of 30-year TIPS, issued in August 2020 at a negative yield of -0.272%, the first time in history that TIPS sold at auction at a negative yield🍾🥂. Crazy times, back in mid-2020, as the Fed had just gotten done with $3 trillion in QE over a three-month period. The TIPS mature in February 2050, CUSIP Number 912810SM1.
The Fed, back when it still did QE, waded deeply into the relatively small TIPS market, mostly in the secondary market through its Primary Dealers. At this TIPS auction in August 2020, the New York Fed’s SOMA desk bought $608 million of these 30-year TIPS, or about 9% of the total issued.
At 58.78% of par value, bought back $40 million of 30-year TIPS, issued in August 2021 at a negative yield of -0.292%, maturing in February 2051, CUSIP Number 912810SV1.
The Fed bought $1 billion of these 30-year TIPS, or about 11% of the total issued.
At 91.47% of par value, bought back $15 million of 10-year TIPS, issued in September 2022 at a yield of 1.248%, maturing in July 2032, CUSIP Number 91282CEZ0. In September 2022, the Fed had already started QT and didn’t buy any TIPS at that auction.
At 96.0% of par value, bought back $40 million of 10-year TIPS, issued in July 2023 at a yield of 1.49%, maturing in July 2033, CUSIP Number 91282CHP9. At this auction in July 2023, the Fed was deep into QT and bought none.
On October 16: bought back 7-year & 10-year notes, including:
At 90.48% of par value, bought back $175 million of 7-year notes, issued in October 2020, at a yield of 0.60%, maturing in October 2027, CUSIP Number 91282CAU5. These 7-year notes have 3 years left to run and so they currently trade like 3-year notes.
At this auction in October 2020, when QE was in full swing, the Fed bought $5.6 billion, or 9.6% of the total.
At 96.55% of par value, bought back $176 million of 10-year notes, issued in February 2018 at a yield of 2.81%, maturing in February 2028, CUSIP Number 9128283W8. At the auction in 2018, the Fed bought $4.5 billion or 19% of the total.
At 90.3% of par value, bought back $139 million of 7-year notes, issued in December 2020 at a yield of 0.66%, maturing in December 2027, CUSIP Number 91282CBB6. At the auction in 2020, the Fed bought $9.0 billion of them, or about 13% of the total.
At 90.25% of par value, bought back $129 million of 7-year notes, issued in August 2021 at a yield of 1.16%, maturing in August 2028, CUSIP Number 91282CCV1. At the auction in August 2021, the Fed bought $7.8 billion of them, or 11.2% of the total.
On October 10: bought back 7-year & 10-year notes, including:
At 100.25% of par value, bought back $17 million of 7-year notes, issued in October 2022 at a yield of 4.03%, maturing in October 2029, CUSIP Number 91282CFT3.
These notes have a coupon interest rate of 4.0%, and with 5 years left to run, they trade like a 5-year yield which is currently 4.03%. At the auction in October 2022, the Fed didn’t buy any of them.
At 88.29% of par value, bought back $377 million of 10-year notes, issued in April 2020 at a yield of 0.78%, maturing in February 2030, CUSIP Number 912828Z94.
It seems hard to imagine today that someone would actually buy a 10-year note with a yield of 0.78%, but the March-to-May 2020 period was when the Fed was buying about $3 trillion in securities in the secondary market through its primary dealers in order to drive down yields and digest the $3 trillion in new Treasuries that the government was issuing over the same period.
The bulk of QE – to increase its holdings – was done in the secondary market. So at this auction in April 2020, the Fed only bought $82 billion of the 7-year notes.
At 82.71% of par value, $390 million of 10-year notes, issued in August 2020 at a yield of 0.677%, which was the lowest yield ever for a 10-year note sold at auction🍾🥂. They mature in August 2030, CUSIP Number 91282CAE1.
At the auction in August 2020, the Fed bought $22.6 billion of these misbegotten 10-year notes, or 37% of the total – and these duds are lodged on it balance sheet to this day. The Fed will hold them to maturity, collect 0.625% in coupon interest a year, and get face value back at maturity.
On October 2: bought back TIPS, including:
At 97.08% of par value, $22 million of 10-year TIPS, issued in January 2018 at a yield of 0.548%, maturing in January 2028, CUSIP Number 9128283R9.
At the auction in January 2018, the Fed bought $1.9 billion of these TIPS, or about 12.8% of the total.
At 99.28% of par value, $10 million of 5-year TIPS issued in June 2023 at a yield of 1.83%, maturing in April 2028, CUSIP Number 91282CGW5. At that auction in June 2023, the Fed bought none.
Some buyback technicalities.
Buybacks are not new. The Treasury Department was buying back older Treasury securities in 2000 through 2002, and on a minuscule scale once or twice a year from 2014 onward, with amounts such as $25 million a year, just to test the plumbing.
Treasury cites some reasons for the buybacks, such as to help manage its cash after tax season when it’s drowning in cash; and creating liquidity in an end of the market that lacks it. The effect – and the real reason for them – would be slightly lower yields in that end of the market.
Buybacks are supposed to ease a liquidity problem. The problem in the Treasury market is that older “off-the-run” Treasury securities are hard to sell because there is little trading in them, and sellers have to accept a (small-ish) haircut when they do find a buyer. The US Treasury market is the most liquid bond market in the world, but it’s not that liquid for older Treasury securities. So Treasury stepped in as buyer of “off-the-run” securities. Recently issued “on-the-run” securities, however, are easy to sell without haircut.
Buybacks are not QE. QE involved money creation – buying bonds with newly created money. That’s what the Fed did until 2022. Treasury cannot create money. It can only get money from collecting taxes and from borrowing – huge amounts of borrowing – which means that the Treasury Department issues new securities that add to the debt; the cash proceeds get spent on budget items and servicing the existing debt, and a small portion of those cash proceeds is now used to buy back off-the-run Treasury securities. Treasury cancels the securities it bought back and they cease to exist. In other words, with these buybacks, Treasury is swapping older bonds for brand new bonds.
Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the beer and iced-tea mug to find out how:
Would you like to be notified via email when WOLF STREET publishes a new article? Sign up here.
Help me understand the reasoning of the Treasury department.
You sell some suckers bonds that pay far less interest than any reasonable person would ever hope to receive, and you then buy them out a few years later so you can issue bonds that pay way more interest? All to try to drive down long term bond rates when the Fed has turned on QT?
What am I not understanding? This seems like the stupidest thing ever.
Happy2 help out, so to speak.
They buy them back at a big discount, as I pointed out, for example at 88 cents on the dollar. So they lower the debt amount, but pay a higher interest on that smaller amount. In terms of interest expense, it should be a wash (the market works that way, no free lunches). But the overall debt is lower.
What it does accomplish is pushing down the yields a little in that end of the Treasury market because now there’s an additional buyer of hard to sell “off-the-run” Treasury securities.
Shouldn’t the price be lower than 88 cents on the dollar? The 10 year yield is about 4.2% now, the treasury sold 2020 was 0.677%. That’s about a 3.5% difference annually. 6 years left on the bond would roughly equate 3.5% times 6, that’s about 20% compared to the 12% discount they were bought back at. Is it safe to say the Treasury loses out on that difference? It seems like an act of mercy after screwing bond holders so bad!
And why the hell would folks buy long term bonds at low and negative rates? Did the expect the rates to go even more negative? If I recall SV bought some of those long term bonds and it led to that mess? You’d think decision makers at these large financial institutions have some math/prediction skills!
Maybe I’m missing something and that’s why I’m a lowly scientist/engineer.
I was thinking that too. There should be a much bigger discount to par with such a low coupon.
IIRC when I was buying 8-10 month t-bills I was usually paying around $0.90 on the dollar. Bills being zero coupon of course.
In terms of your first paragraph: You can’t figure bond prices that way in your head. It’s a very complex calculation involving dates. Back in the mid-1980s, we used to do it with our HP 12c calculators and date tables. Now algos do that. This 10-year note trades today like a 5-year-5-month note because it only has 5 years and 5 months left to run. The loss diminishes each day the closer the note gets to maturity date. Just before the maturity date, the loss will be zero, and the note will be worth face value plus 6 months of interest, which is what the holder will get paid.
In terms of your second paragraph: yes, at the time all the big bond gurus were predicting that the 10-year yield would go negative at the auction, as it had already done in Europe. Their hope was to make money on the price as the 10-year yield sinks deeper and deeper into the negative, which has turned out to be the miscalculation of the century, and three regional banks already got taken out and shot because of it.
Hi, help me understand.
“At 58.78% of par value, bought back $40 million of 30-year TIPS”
What is the accounting for that? My understanding is government debt is recorded at par, so there used to be $40m of debt. It only cost $23.5 to retire that debt. They then issue $23.5m of new debt at a par value of $23.5m to pay for that.
So that leaves a profit of $16.5m to reduce the deficit this year.
Why is this not being done at scale to reduce debt as a percentage of GDP to lower levels? It doesn’t resolve the interest payments, the interest payments are a wash, but the lower debt as a percentage of GDP should result in a credit upgrade and lower yields?
1. Correct, it did reduce the debt.
2. But no, it doesn’t create a “profit” because the government isn’t a for-profit enterprise, and doesn’t do profit/loss accounting.
3. The government is paying a higher percentage interest rate on the buyback amount (borrowed at higher current rates), but it’s paying the higher percentage rate on a smaller dollar-debt amount, and so the interest expense in dollars should come out about the same (that’s how bonds are priced, markets work that way, no free lunch).
4. Because the dollar-interest expense is the same (see #3), this action has no impact on the credit rating of the US. Credit ratings rate the ability of the US to service its debts and the probability of it not being able to do so.
Thanks, I’m an accountant so the debits and credits have to balance to get my head around things. Government accounting is something of a mystery.
There are some interesting charts here looking at market value vs par value of national debt as a share of gdp. Back in October 2023 the market value of debt as a share of GDP got back to pre-pandemic levels. Par value was 33.7tn, market value 30.2tn, a 10% gap. With long term interest rates climbing again this level of difference may happen again soon.
https://www.dallasfed.org/research/econdata/govdebt#charts
I understand there is no free lunch, but given the focus on the particular statistic of debt as a share of gdp by the press and ratings agencies it would seem to still be beneficial to bring down that percentage? No free lunch, but also no real cost to do so, and a real possible benefit of a ratings upgrade, or at least avoid another downgrade, resulting in lower yields?
I don’t think that’s a good idea. The big risk with the government buying back its own bonds in a large way is two-fold:
1. It would have to issue even more new bonds, in the amount of the buybacks. So if it wants to buy back $100 billion a month at market value, it would have to issue an additional $100 billion in bonds a month, on top of the huge issuance already taking place. I would nearly double the new issuance. Maybe that’s getting kind of scary here.
2. If a large buyer is buying large amounts of illiquid bonds, those bond prices are going to jump a lot (yields go down), which means that the government is overpaying for these bonds. It needs to buy them back at big discounts for this to make sense. But if the discounts vanish because of its hefty buying, it would be shooting itself in the foot by overpaying for these buybacks.
Are the Treasury and the Federal Reserve working this in tandem? Or are they working this independently, as they claim?
The Fed has ZERO to do with these buybacks.
In fact, the Treasury is doing this to counteract the Fed’s monetary policy. The government has done everything it can to manipulate DOWN long-term yields over the past two years, while the Fed has tried to PUSH UP long-term yields with QT and higher rates to crack down on inflation. They’re essentially fighting each other.
And there was me thinking it was just a work around for the government debt limit. The Treasury redeems $1.00 of par value debt for $0.88, which reduces the debt with respect to the limit by $1.00. It then issues $1.00 of par value debt for $1.00, at a higher interest rate, which brings the debt with respect to the limit back to where it was originally, but with $0.12 in hand.
That’s nonsense. I don’t know how you people can twist everything around like this.
The Treasury issues new Treasuries all the time to raise cash, and some of that cash went to buy back old debt at 88 cents on the dollar, and so the buyback had the effect that the debt fell by the amount of the discount.
And the amounts are way too small to make any difference in terms of any debt limit ($7 billion a month, of $35.8 TRILLION in debt. It’s not even a rounding error.
My whole point, the debt fell by the amount of the discount, which means, with respect to the debt limit, that new debt can be raised by the amount of the discount. I accept that the amounts may be insignificant.
This is such a good idea, why don’t they scale up and trade in size?
By issuing bonds when rates are low (recessions), and buying them back when rates are high (periods of growth), they could finance a good part of the deficit – or maybe even use the trading profits to amortize the debt!
Call it Keynesian Bond Trading…
It can buy bonds only by issuing new bonds to raise the cash to buy back the old bonds, dollar for dollar. Those two always go together. The government is not sitting on a Strategic Cash Reserve (SCR) or whatever that it can deploy to buy back bonds.
But what it can do and should do, is issue lots of long-term debt when rates are low, and issue more T-bills when rates are high, and then refinance the T-bills with long-term debt when rates are low again.
The Yellen Treasury failed to issue lots of long-term bonds when rates were low in 2020. But in a way, that was a good thing, because if it had issued lots more of this long-term stuff, it would have ended up on our banks’ balance sheets, and our banking system would have been toast by now. Maybe they figured this much in 2020 and tried to not blow up the banking system in the future.
Wolf, help me understand something:
/Why/ are off-the-run coupons harder to sell? Don’t they just trade at enough of a discount that the effective yield-till-maturity is the same APY as a newer coupon?
What’s the different between a 10Y note with a half % coupon that trades at a big discount, and a zero coupon bond? Do investors/traders not like zero coupons with that much duration?
Seems like the government is bailing out some of these bonds that are worth way less then what they were bought for… Another form of yield control.
Gonna be great when these clowns lose control on
the curve.
It’ll actually be terrible and probably worse then the GFC but that’s the bed our leadership has made