He used Bennett twice in the 3rd graf, but further down he used Bessent. I would guess that if AI made the error it would have made it consistently. I am guessing human.
"It’s not a new idea. Historically, perpetual government debt was more common than fixed maturity debt precisely because you didn’t have to roll it over. The Downton Abbey set held perpetuities."
What a brilliant statement! Perhaps it should have been the title.
From my days teaching money & banking: “Asset A is more liquid than asset B if A can be converted into money at a lower cost than can B.” Of course, that’s for two individual assets. Expanding the definition to the market level, we can talk about asset groups (industries, risk rating, etc.)
Now, if only we could get the impediment of vested interest in the existing privileged trading houses in Treasuries out of the way ... Who wouldn't like to be one of them?
There is something to this. Treasurys should also be openly traded, not phone calls to big banks, who "provide liquidity" when it suits them and go screaming to the Fed when it does not. Also, the current system means the big banks earn the bid/ask spread on the entire $40 trillion every time it rolls over. Perpetuities eliminate that.
I like the idea of simplification and of avoiding the off run problem. But I think that having 2 different types of perpetuals would create too much confusion. I am not sure how you keep the coupons synchronized. I also think that having securities with fixed maturities allows entities such as pension plans to engage in liability matching.
My suggestion is that we issue perpetuals with a fixed coupon (interest only securities IOs and fixed term non interest paying principal only bonds POs. The POs would intially be issued with a spectrum of maturities say 1 to 10 years. later on they could be issued in the form of 10 year only bonds. but there would always be a slug of bonds with shorter maturities in the market. For inflation protection the government could issue put options that would pay a premium if tendered with a maturing PO or pay a face amount for a perpetual. There would be a trading market for the options that would give us a very sensitive reading about market expectations for inflation.
A perpetuity is a rather simple security to describe. And, providing an quasi-inflation hedge is also simple to arrange.
Suppose, the perpetuity is to have an issue price of $1000.00 per unit. Suppose that the annual coupon is one-one hundreth of a troy ounce of fine 999.9 gold per unit. Furthermore, suppose that the gold price on the date of issue is $4,500.00 per troy ounce of fine 999.9 gold. Doing the algebra, the initial yield in dollars is 4.5% per annum. This is somewhat of a discount to the 30-year Treasury bond today. But, the annual coupon is based on the price of gold at, say, the London fixing on a specific day at noon GMT in a specific month---say, on the second Wednesday of the month of December. Additionally, the obligation is settled in the equivalent fine 999.9 value of gold in U.S. dollars rather than in fine 999.9 gold.
This arrangement gives you the perpetuity you seek along with a quasi-inflation hedge. If the issue is large enough, it will also incentivise the Treasury and the government to seek a low inflation commensurate with economic growth and full employment, almost surely.
There is no restriction on the Treasury calling the issue for redemption, except it is not permitted to call the issue for redemption earlier than the 10th year after original issue. There is no pre-specified call price.
Additional issues of the perpetuity can be floated at the sole discretion of the Treasury, subject to restrictions on issue size and creditworthiness, etc.
The original issue and subsequent issues, if any, are not retractable.
The market value of the perpetuity will vary with the gold price of the dollar.
Just for clarity from whence the variable rate that keeps the perpetuity at par? A daily market for the V perps akin to the commercial paper or money market fund but for V perps?
The Treasury can pay the Fed's interest on excess reserves, and allow the quantity relative to 1-3 month bills to fluctuate. Or the Treasury can have a daily auction of a fixed quantity, auctioning the rate to be paid that night at a price of 1.000
I am usually a big fan of your work and love to recommend it to academics and practicioners. You are one of the (very few) economists I genuinely like, but I have to admit that this writing was a bit off-putting to me. I am working as a market maker myself, so I hope my thoughts here are not completely out-of-the-blue.
First of all, the sentence "Gemini made this graph for me which looks right but I can't vouch for the source" is not the standard I expect from someone like you. I am sure you have access to live market data where you can pull that data and create that chart in minutes. In my opinion you are jeopardizing factual correctness and integrity with something like this.
Second of all, there are a few market related problems I have with your proposition.
The implication of "lower-liquidity" -> "lower prices/higher yield" is not always correct and shouldn't be a general assumption. There is ambiguity in the market (I know examples from rates & credit where the relationship is inverse).
Furthermore, the idea of the treasury publishing a "daily yield curve" fit to the on-the-runs and buying/selling if a bond is out of line sounds insane to me. The treasury would here act as a market maker (always providing bids/offers), which I am not sure it should. The problem of "what is OTR and OffTR" would appear and attract speculators/inefficiencies. The treasury would also basically just set their own curve (providing you also assume infinite firepower to buy/sell bonds, ignoring positive/negative/size balance sheet implications here). If you do not assume infinite firepower, I am sure that the market loves to test the limits.
When you wrote "the Treasuy can buy, at a discount [...], smart management" - where would the Treasuy get the money for this? From shorting the other bond? That basically amounts to increasing the outstanding volume of the bond the Treasury has just shorted (traders call this a "Tap"). But knowing that mechanism would just lead to market participants do a basis-trade equivalent..
Also, imagine a scenario of very bad news in the US, people start selling Treasuries. "Normal" market makers (investment banks, trading houses) now can completely stop buying bonds since they know the Treasury is there to buy them from panicky investors. Now the Treasury is buying probably billions of bonds at the daily mark, which at this point is probably too good of a price. Tomorrow the Treasury has to remark those yields higher (more sellers than buyers), generating a huge mark-to-market loss on their own balance sheet (I guess it is arguable if that matters - but I'm sure journalists would love to make that a story) or perhaps not marking it wide enough and completely ruining market price setting. Making the daily mark instantenuous/real-time to avoid this? Not sure if this works without getting rid of free market dynamics.
If I misunderstood you and you are meaning to have a very wide interval where the Treasury is buying/selling, normal market makers would always price within these limits to generate flow, so the whole exercise is rendered useless (especially in a highly liquid market like treasuries).
Off the run bonds are btw also needed to mark proper interpolated yield curves. If you would always just have OTR bonds (1,2,3,5,10,20,30Y for example), the space between them is more prone to error. Maybe there is a good reason for the 12.7Y bond to trade wider? The current "stability" of the curve is given by "speculators" (per definition not arbitrageurs) who take on exactly the curve risk to often "smooth things out".
Your proposal to just provide a fixed-coupon and a floating rate perpetuity also runs into many errors in my opinion. First of all, it complete ignores the vast surrounding market ecosystem that is built on the treasury curve that acts as a benchmark for nearly every fixed income security and derivative out there.
I know you briefly wrote "It can also offer fixed-for-floating swaps". Wouldn't that also create a curve? How would the Treasury determine the fair price of those derivatives - How would that fit into the Treasries' raison d'etre (same argument to the daily marks above)? Would it still be market dynamics like right now? Would they act as a market maker and get paid for the service? Does that align with incentives? It would also complicate the life of every Asset-Liability-matching investor out there that curerntly relies on bonds with different maturities. Issuing to "fill the demands so it can borrow at lower rates", thus creating again inefficiencies (why is there a fixed maturity bond now that differs in return from the fixed-floating-swap they provde? Is that not the same sitation we tried to get rid of?
I also doubt that companies would get the privilege of issuing solely perpetual debt, so them issuing fixed-maturity debt (like right now) would not only get more complicated, but also imply a treasury curve in itself (Imagine a corporate bond from company A. People would still want to receive rates risk + credit spread. The market would distill the credit spread (since these bonds trade on that) and voila, you have a US interest curve again and the tail wags the dog).
A fixed coupon perp would also have the highest convexity, potentially making the market a bit more unstable and more prone to non-linear risk.
Finally to end, I am incredibly happy about you adding "[...] and I know for sure there is no theory or objective measurement of any of them)" regarding liquidity. Liquidity is a difficult story, and here we are talking about one of the most liquid assets out there. I also generally like the idea of providing a facility for everyone to receive a "Fed reserve rate", think there are many positives regarding that (But I have never fleshed out my thoughts on that). But there will always be a direct or indirect "US risk maturity curve" as long as time is a necessary variable to calculate any investment return.
Maybe I have misunderstood some things which are easily corrected and I am also sure I have swapped the word "Treasury" & "Fed" at least once here - I apologize for that.
I would be incredibly grateful for any reply to this. No AI was used in writing this. I like the trade of having humanity in my text at the cost of less polished textwork.
With best regards from usually big fan of your work!
Too much here for point by point. Today, people want fixed-value floating-rate debt. The treasury doesn't offer it, so private intermediaries, mutual funds, turn A into B. People want strips. Treasury doesn't offer strips, so intermediaries turn bonds into strips. You posit that people want fixed maturity coupon bonds. OK, but why must the treasury provide them directly? Easy enough for the same banks who create strips to buy short and long perpetuities, and issue coupon bonds. They ought to be better at matching such detailed customer demand than Treasury is.
I agree with the person on most of their point for point observations -
I ran Treasury trading at one of the top 5 dealers for a number of years. Innovation has often been welcome and successful. TIPs, STRIPS, and other structures and issuing techniques has all been welcome additions. The result is the interest rate market structure is extremely durable today. Collectively, it has evolved into the intersection of cash Treasuries, rate swaps, futures contracts, repos and MBS TBAs.
The liquidity and resiliency of these markets are highly integrated so that transactional needs in one part of the market are reliably and smoothly distributed
throughout the entire liquidity base.
As a result, temporary pricing dislocations, when they happen, are very small and quickly addressed by an army of high speed electronic trading systems and arbitrageurs.
Off the run “illiquidity” for longer dated bonds is only a theoretical issue - since even during the most stressful periods these so-called illiquid bonds are still far more liquid than virtually other benchmark financial instruments. In fact, it is more common to talk about the on the run treasuries having a significant “high liquidity” premium, instead of an “illiquidity discount” for the off the run bonds.
I very much enjoy your commentaries, so keep it going. The core problem for the US Treasury is issuing too much debt and your suggestion that the government consider innovative ways to camouflage this problem with financial engineering is a sober wake-up call.
There would be pretty massive regulatory implications to moving to fixed-coupon perpetuities in accounting wouldn't there - wouldn't it be effectively impossible for banks to hold them "at par" without ruining the whole "1 CUSIP to rule them all" liquidity concept? Perhaps a feature from an economist's perspective, but wouldn't this massively diminish demand for the instruments? I imagine this is the main hurdle to overcome.
FOMC gave the directive to NY Fed in Dec 2025 to halt all buying of long term bonds. Only T-bill purchases are authorised now for reserve purposes. So, there's no chance Bessent called Warsh, and independence was never an issue. Just good old fashion debt management.
The Secretary of the Treasury is not “Bennett.” AI editor?
He used Bennett twice in the 3rd graf, but further down he used Bessent. I would guess that if AI made the error it would have made it consistently. I am guessing human.
Autocorrect, but that does not absolve me from proofreading.
"It’s not a new idea. Historically, perpetual government debt was more common than fixed maturity debt precisely because you didn’t have to roll it over. The Downton Abbey set held perpetuities."
What a brilliant statement! Perhaps it should have been the title.
Well done John!
From my days teaching money & banking: “Asset A is more liquid than asset B if A can be converted into money at a lower cost than can B.” Of course, that’s for two individual assets. Expanding the definition to the market level, we can talk about asset groups (industries, risk rating, etc.)
Now, if only we could get the impediment of vested interest in the existing privileged trading houses in Treasuries out of the way ... Who wouldn't like to be one of them?
There is something to this. Treasurys should also be openly traded, not phone calls to big banks, who "provide liquidity" when it suits them and go screaming to the Fed when it does not. Also, the current system means the big banks earn the bid/ask spread on the entire $40 trillion every time it rolls over. Perpetuities eliminate that.
I like the idea of simplification and of avoiding the off run problem. But I think that having 2 different types of perpetuals would create too much confusion. I am not sure how you keep the coupons synchronized. I also think that having securities with fixed maturities allows entities such as pension plans to engage in liability matching.
My suggestion is that we issue perpetuals with a fixed coupon (interest only securities IOs and fixed term non interest paying principal only bonds POs. The POs would intially be issued with a spectrum of maturities say 1 to 10 years. later on they could be issued in the form of 10 year only bonds. but there would always be a slug of bonds with shorter maturities in the market. For inflation protection the government could issue put options that would pay a premium if tendered with a maturing PO or pay a face amount for a perpetual. There would be a trading market for the options that would give us a very sensitive reading about market expectations for inflation.
A perpetuity is a rather simple security to describe. And, providing an quasi-inflation hedge is also simple to arrange.
Suppose, the perpetuity is to have an issue price of $1000.00 per unit. Suppose that the annual coupon is one-one hundreth of a troy ounce of fine 999.9 gold per unit. Furthermore, suppose that the gold price on the date of issue is $4,500.00 per troy ounce of fine 999.9 gold. Doing the algebra, the initial yield in dollars is 4.5% per annum. This is somewhat of a discount to the 30-year Treasury bond today. But, the annual coupon is based on the price of gold at, say, the London fixing on a specific day at noon GMT in a specific month---say, on the second Wednesday of the month of December. Additionally, the obligation is settled in the equivalent fine 999.9 value of gold in U.S. dollars rather than in fine 999.9 gold.
This arrangement gives you the perpetuity you seek along with a quasi-inflation hedge. If the issue is large enough, it will also incentivise the Treasury and the government to seek a low inflation commensurate with economic growth and full employment, almost surely.
There is no restriction on the Treasury calling the issue for redemption, except it is not permitted to call the issue for redemption earlier than the 10th year after original issue. There is no pre-specified call price.
Additional issues of the perpetuity can be floated at the sole discretion of the Treasury, subject to restrictions on issue size and creditworthiness, etc.
The original issue and subsequent issues, if any, are not retractable.
The market value of the perpetuity will vary with the gold price of the dollar.
Just for clarity from whence the variable rate that keeps the perpetuity at par? A daily market for the V perps akin to the commercial paper or money market fund but for V perps?
The Treasury can pay the Fed's interest on excess reserves, and allow the quantity relative to 1-3 month bills to fluctuate. Or the Treasury can have a daily auction of a fixed quantity, auctioning the rate to be paid that night at a price of 1.000
Thx
John needs to speak to someone who knows something about the bond market. Problem is simple: markets work government deficits are way too high.
Dear Professor,
I am usually a big fan of your work and love to recommend it to academics and practicioners. You are one of the (very few) economists I genuinely like, but I have to admit that this writing was a bit off-putting to me. I am working as a market maker myself, so I hope my thoughts here are not completely out-of-the-blue.
First of all, the sentence "Gemini made this graph for me which looks right but I can't vouch for the source" is not the standard I expect from someone like you. I am sure you have access to live market data where you can pull that data and create that chart in minutes. In my opinion you are jeopardizing factual correctness and integrity with something like this.
Second of all, there are a few market related problems I have with your proposition.
The implication of "lower-liquidity" -> "lower prices/higher yield" is not always correct and shouldn't be a general assumption. There is ambiguity in the market (I know examples from rates & credit where the relationship is inverse).
Furthermore, the idea of the treasury publishing a "daily yield curve" fit to the on-the-runs and buying/selling if a bond is out of line sounds insane to me. The treasury would here act as a market maker (always providing bids/offers), which I am not sure it should. The problem of "what is OTR and OffTR" would appear and attract speculators/inefficiencies. The treasury would also basically just set their own curve (providing you also assume infinite firepower to buy/sell bonds, ignoring positive/negative/size balance sheet implications here). If you do not assume infinite firepower, I am sure that the market loves to test the limits.
When you wrote "the Treasuy can buy, at a discount [...], smart management" - where would the Treasuy get the money for this? From shorting the other bond? That basically amounts to increasing the outstanding volume of the bond the Treasury has just shorted (traders call this a "Tap"). But knowing that mechanism would just lead to market participants do a basis-trade equivalent..
Also, imagine a scenario of very bad news in the US, people start selling Treasuries. "Normal" market makers (investment banks, trading houses) now can completely stop buying bonds since they know the Treasury is there to buy them from panicky investors. Now the Treasury is buying probably billions of bonds at the daily mark, which at this point is probably too good of a price. Tomorrow the Treasury has to remark those yields higher (more sellers than buyers), generating a huge mark-to-market loss on their own balance sheet (I guess it is arguable if that matters - but I'm sure journalists would love to make that a story) or perhaps not marking it wide enough and completely ruining market price setting. Making the daily mark instantenuous/real-time to avoid this? Not sure if this works without getting rid of free market dynamics.
If I misunderstood you and you are meaning to have a very wide interval where the Treasury is buying/selling, normal market makers would always price within these limits to generate flow, so the whole exercise is rendered useless (especially in a highly liquid market like treasuries).
Off the run bonds are btw also needed to mark proper interpolated yield curves. If you would always just have OTR bonds (1,2,3,5,10,20,30Y for example), the space between them is more prone to error. Maybe there is a good reason for the 12.7Y bond to trade wider? The current "stability" of the curve is given by "speculators" (per definition not arbitrageurs) who take on exactly the curve risk to often "smooth things out".
Your proposal to just provide a fixed-coupon and a floating rate perpetuity also runs into many errors in my opinion. First of all, it complete ignores the vast surrounding market ecosystem that is built on the treasury curve that acts as a benchmark for nearly every fixed income security and derivative out there.
I know you briefly wrote "It can also offer fixed-for-floating swaps". Wouldn't that also create a curve? How would the Treasury determine the fair price of those derivatives - How would that fit into the Treasries' raison d'etre (same argument to the daily marks above)? Would it still be market dynamics like right now? Would they act as a market maker and get paid for the service? Does that align with incentives? It would also complicate the life of every Asset-Liability-matching investor out there that curerntly relies on bonds with different maturities. Issuing to "fill the demands so it can borrow at lower rates", thus creating again inefficiencies (why is there a fixed maturity bond now that differs in return from the fixed-floating-swap they provde? Is that not the same sitation we tried to get rid of?
I also doubt that companies would get the privilege of issuing solely perpetual debt, so them issuing fixed-maturity debt (like right now) would not only get more complicated, but also imply a treasury curve in itself (Imagine a corporate bond from company A. People would still want to receive rates risk + credit spread. The market would distill the credit spread (since these bonds trade on that) and voila, you have a US interest curve again and the tail wags the dog).
A fixed coupon perp would also have the highest convexity, potentially making the market a bit more unstable and more prone to non-linear risk.
Finally to end, I am incredibly happy about you adding "[...] and I know for sure there is no theory or objective measurement of any of them)" regarding liquidity. Liquidity is a difficult story, and here we are talking about one of the most liquid assets out there. I also generally like the idea of providing a facility for everyone to receive a "Fed reserve rate", think there are many positives regarding that (But I have never fleshed out my thoughts on that). But there will always be a direct or indirect "US risk maturity curve" as long as time is a necessary variable to calculate any investment return.
Maybe I have misunderstood some things which are easily corrected and I am also sure I have swapped the word "Treasury" & "Fed" at least once here - I apologize for that.
I would be incredibly grateful for any reply to this. No AI was used in writing this. I like the trade of having humanity in my text at the cost of less polished textwork.
With best regards from usually big fan of your work!
Too much here for point by point. Today, people want fixed-value floating-rate debt. The treasury doesn't offer it, so private intermediaries, mutual funds, turn A into B. People want strips. Treasury doesn't offer strips, so intermediaries turn bonds into strips. You posit that people want fixed maturity coupon bonds. OK, but why must the treasury provide them directly? Easy enough for the same banks who create strips to buy short and long perpetuities, and issue coupon bonds. They ought to be better at matching such detailed customer demand than Treasury is.
John,
I agree with the person on most of their point for point observations -
I ran Treasury trading at one of the top 5 dealers for a number of years. Innovation has often been welcome and successful. TIPs, STRIPS, and other structures and issuing techniques has all been welcome additions. The result is the interest rate market structure is extremely durable today. Collectively, it has evolved into the intersection of cash Treasuries, rate swaps, futures contracts, repos and MBS TBAs.
The liquidity and resiliency of these markets are highly integrated so that transactional needs in one part of the market are reliably and smoothly distributed
throughout the entire liquidity base.
As a result, temporary pricing dislocations, when they happen, are very small and quickly addressed by an army of high speed electronic trading systems and arbitrageurs.
Off the run “illiquidity” for longer dated bonds is only a theoretical issue - since even during the most stressful periods these so-called illiquid bonds are still far more liquid than virtually other benchmark financial instruments. In fact, it is more common to talk about the on the run treasuries having a significant “high liquidity” premium, instead of an “illiquidity discount” for the off the run bonds.
I very much enjoy your commentaries, so keep it going. The core problem for the US Treasury is issuing too much debt and your suggestion that the government consider innovative ways to camouflage this problem with financial engineering is a sober wake-up call.
There would be pretty massive regulatory implications to moving to fixed-coupon perpetuities in accounting wouldn't there - wouldn't it be effectively impossible for banks to hold them "at par" without ruining the whole "1 CUSIP to rule them all" liquidity concept? Perhaps a feature from an economist's perspective, but wouldn't this massively diminish demand for the instruments? I imagine this is the main hurdle to overcome.
FOMC gave the directive to NY Fed in Dec 2025 to halt all buying of long term bonds. Only T-bill purchases are authorised now for reserve purposes. So, there's no chance Bessent called Warsh, and independence was never an issue. Just good old fashion debt management.