Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Treasuries are amazing. The US government bond market is so big, liquid, safe and widely-acceptable as quasi-money that it commands something that economists call a “convenience yield”.
In reality, it functions more like a premium — the real-world, semi-measurable manifestation of what is often otherwise referred to as America’s “exorbitant privilege”.
Treasuries are just so darn attractive to hold and easy to use for a wide variety of purposes that the US government can actually borrow more cheaply than even the genuine risk-free rate.
A few years ago the NY Fed estimated the savings at ca 20-40 basis points a year, or ca $35bn a year over the past two decades. Since 2020 it has amounted to about $70bn. Which may seem piddling in the context of a $28.6tn market, but still amounts to more than three USAIDs, two NASAs or one Department of Justice (with change left for an SEC and CFTC).
Unfortunately, it looks like the convenience yield is another American institution getting butchered in 2025. From a newish paper from NYU finance professors Viral Acharya and Toomas Laarits, with FTAV’s emphasis in bold:
This note explains how the “tariff war” shock of early April 2025 affected the convenience yield of US Treasuries. Its erosion at the long end is consistent with a reduction in the safe-asset hedging property of long bonds, reflected in a rising stock-bond covariance computed using intraday data. Decomposing the Treasury yield into the risk-free rate, credit spread, and the convenience yield components reveals that it is covariance due to the convenience yield component that rose for long bonds. The same decomposition reveals that the short end of the Treasury curve continued to exhibit the safe-asset hedging property both due to a lowering of the (expected) risk-free rates as well as an increase in the convenience yield component. These effects are consistent with stagflation risk, withdrawal of safe-asset investors, rotation towards shorter-term Treasuries and gold, and unwinding of cash-futures basis trades.
In plainer language, short-term Treasury bills still seem to boast some of the precious convenience premium, but longer term Treasury bonds have lost some of their specialness.
The paper is from early May, when the tariff shock was still fairly fresh, so the authors stress that their finding of a “partial but not complete reversal” in the convenience yield is still a preliminary one.
Acharya and Laarits primarily use the yield differential between conventional and inflation-linked Treasuries to estimate the convenience premium. The former consistently trades at higher prices even once you factor in market expectations for inflation that are derived from derivative prices, and this gap is a decent proxy for the convenience yield (the NY Fed compared Treasury yields to the option-derived risk-free “box rate” instead, which is somehow more aesthetically pleasing, if a lot more complicated).
In a separate but clearly related paper, Acharya and Laarits look at when the convenience yield is particularly high. Unsurprisingly they find that investors are particularly willing to pony up a premium for Treasuries when they perform well as a shock absorber. In other words, when Treasuries rally hard as stocks puke, the convenience of US government bonds that soften the blow is understandably super high. Conversely, things that make investors a bit wary of the US — such as debt ceiling shenanigans — erodes the convenience yield.
However, a separate paper — this one by Zhengyang Jiang, Robert Richmond and Tony Zhang, of Northwestern, NYU and the Federal Reserve respectively — takes a slightly more granular look at what bond maturities command the greatest convenience yield.
Interestingly (and if you’ve gotten this far in the post we assume you find this as interesting as we do), Jiang, Richmond and Zhang’s main conclusion is very different: they estimate that the convenience yield has actually been falling for over a decade and is now in many cases negative.
By taking a more granular look at maturities, they conclude that this is because of massive amounts of long-term Treasury sales in recent years. Alphaville’s emphasis below:
Our instrumental variable regression estimates show that increases in Treasury supply cause convenience yields to decline, particularly for long-term debt. An increase in the supply of long-term debt by 5 percent of GDP results in a 0.94 percentage point decline in convenience yields at the 10-year horizon. In contrast, the convenience yield on short-term Treasurys is relatively insensitive to changes in supply.
To understand the aggregate implications of these estimates, we sum up the product of Treasury supply and convenience yields across maturities. The data shows that the seigniorage revenue earned from Treasury convenience has declined by 5 to 10 percent of the annual federal interest expense, largely explained by the decline in convenience yields on medium and long-term Treasurys. The seigniorage revenue earned from short-term Treasurys has remained roughly stable as declines in the convenience yield on short-term Treasurys are roughly matched by increases in short-term Treasury supply.
So why do Jiang, Richmond and Zhang in Convenience Lost find a sharply declining convenience yield when researchers at the NY Fed found it mostly oscillating between the 20-40 bps range since 2010?
Without subjecting ourselves to untangling the mathematics of each paper, it is almost certainly due to what exactly they’re measuring.
While the NY Fed paper uses an options-derived method of deriving a “true” risk-free rate that they can compare actual Treasury yields to — and the Acharya and Laarits paper uses inflation-proofed Treasuries — this latest paper uses swaps. And as most Alphaville readers will know, swap spreads can move in mysterious ways.
In summary, financial economists think that the convenience yield has probably fallen. But they don’t agree on why, or over what time period, or whether it is merely eroded or disappeared completely. Hopefully the journey here has been worthwhile anyway.
