The Gravitational Pull of Zero

Published on June 7, 2021 by Free

GSIB High Quality Liquidity Asset (“HQLA”) portfolios are a mechanism through which low rates in the front-end are exerting downward pressure on longer dated yields. Fed QE has filled bank balance sheets with low yielding reserves, and deprived non-banks of any yield at all in the front-end. An unconstrained investor can escape 0% yields by moving along the risk curve to Bitcoin, but GSIBs are confined by Basel III to the most prosaic investments. GSIBs have both limited balance sheet space and HQLA requirements, so they are optimizing their portfolio by exchanging 0.1% yielding reserves for Treasures. Over the past year they have purchased $350b in Treasuries, tilted towards longer dated maturities. In this post we review why GSIBs are purchasing Treasuries and illustrate the scale and maturity profile of those holdings based on recent regulatory filings.

Almost all bank buying of Treasures is from GSIBs. Excludes trading assets

Why are GSIBs Buying Treasuries?

The banking sector has significantly increased its holdings of Treasuries over the past year, but almost all the increase is from GSIBs. Most banks are understandably uninterested in hold assets yielding less than inflation, but those low yields can still be attractive after taking into account regulatory constraints. In prior posts we have shown how Basel III constraints force GSIBs to adjust their liabilities by pushing out institutional deposits, but those same regulations also place constraints their assets.

Bank of America 2021Q1 earnings call

GSIBs are mandated to hold a sizable HQLA portfolio, but there is some degree of freedom in its composition. The highest quality Level I HQLA include reserves, Treasuries, and Treasury reverse repo. When reverse repo rates rose above IOR in 2018, GSIBs began reshuffling their HQLA portfolio out of reserves and into reverse repo. Now that IOR is at 0.1% and reverse repo is at 0%, GSIBs are shuffling their HQLA holdings towards Treasuries. Some are also significantly increasing their holding of Agency MBS, which is a slightly lower quality HQLA (Level 2a).

Note that the focus in this post is entirely on the bank operating subsidiaries (depository institutions), and that there is significant heterogeneity in Treasury holdings even among GSIBs. The increase in Treasury holdings has largely been in the GSIBs with big bank operating subsidiaries- JPM, BAC, C, WFC. These are the banks most impacted by QE, which filled their limited balance sheets with low yielding reserves. The other 4 GSIBs have different business models (SST and BK are focused on custodial services, while GS and MS are focused on investment banking).

GSIBs Have Moved Along the Rates Curve

Bank regulatory filings suggest that the bulk of the increase in Treasury securities has been in the 5 to 15 year bucket. The filings report Treasury holdings but do not offer a clean break down of Treasury holdings by remaining maturity. The closest breakdown provided lumps Treasuries with other securities such as foreign sovereign debt, state and local government debt, and commercial MBS. However, the level holdings of those other securities are little changed over the past year while Treasury holdings surged. The aggregate $350b increase in securities broken down by remaining maturity almost exactly matches the $357b increase in Treasury holdings. This lends confidence to the view that the GSIBs are moving further along the Treasury curve.

Overall change in securities matches the $357b increase in Treasuries

This marks a big shift in bank portfolio management since the crisis, where GSIBs were not very interested in Treasuries (the first chart above shows only $100b in holdings in 2013) and tended to stay within 5 years of maturity. GSIBs are now big buyers of Treasuries, though they still stay away from the very short and very long maturities. Note that a meaningful portion of their portfolio is going be hedged.

GSIB Buying May Not Be The Only Mechanism

Treasuries yields haven’t budged for months despite hot CPI prints, soaring commodity prices, the promise of even more Treasury issuance, and of course an accommodative Fed. Maybe the bond market believes that inflation is transitory. But there are also mechanisms in the financial system that can lead to subdued yields without having to make assumptions on ‘how smart the market is.’ GSIB HQLA portfolios are one channel, but there could easily be other classes of investors (e.g. foreign official or conservative retirement funds) that are bound by similar constraints and forced by the ZLB to move along the duration curve, putting a lid on yields.

When the steady increase in demand for repo financing from 2018 to 2019 pushed repo rates above IOR, GSIBs reallocated their HQLA portfolios into repo and became the marginal lender in that market. They ran out of money in September 2019 and repo rates exploded higher. Treasury coupon issuance going forward is extremely high, so eventually GSIBs and similar investors will also run out of money. That may be when yields finally rise, if the Fed allows it.

11 comments On The Gravitational Pull of Zero

  • If the banks are shifting along the maturity curve in a significant way, doesn’t that also mean that they are fixing in higher probability that their cost of capital in the future won’t exceed whatever rate they are getting on these treasuries? Or at least that the future cost of capital won’t go up so high as to negate the benefits today?


    • A good portion of the Treasures they own will be swapped from fixed to floating – so they will not take interest risk. Their current balance sheet suggests that funding comes from retail deposits, which pay 0% and will likely still pay 0% even if the Fed raises rates. The pass through from Fed funds to deposit rates is really low.

      • Swaps wouldn’t change the (potential) loss for banks in aggregate though (or maybe the private sector as a whole). I probably shouldn’t have used the word “cost of capital” as what you mentioned is definitely true. When the opportunity cost is taken into account, it would still have to mean that banks in aggregate are fixing in higher probability of rates not moving high enough to make the current move into long duration treasuries worth it wouldn’t it?

      • My thoughts:
        if duration from USTs is swapped into floating, I would expect quite some upward pressure on Swap Spreads, right?
        Have we seen this recently – I don’t think so.
        5Y and 10Y Swap Spread has decreased since the middle of Q1 2021.
        So this leaves me wondering, where all the duration risk is kept….

  • So do you think this means longer end rates will tend towards 0 as well? How do you think about the extension of duration in light of an eventual taper/reduction in fed accommodation? Especially if you do get some sort of tantrum reaction, do you see a potential risk negative response to lessened accommodation as a countervailing force to the reduction in accommodation/possible inflation? What do you think this means for intermediate and longer dated yields in the medium term?

  • Thanks for the interesting post.
    Could you add a link to the source data?
    That would be highly appreciated.
    Thank you.

    • Everything I put out has a public source. If you want the underlying data to study yourself I can help you find it – just shoot me an email.

      • Thanks. Interesting post. Could you please point me to the source data that shows the GSIB holdings breakdown by maturity.

        • Every bank must file a call report, which include a breakdown of securities held by maturity by asset class. The call reports can be found here. The relevant section is Schedule RC-B.

          • sir you are a dag-gum wizard! thanks for helping demystify the smoke & mirrors show the fed has been putting on for us. It seems the divergence between you and deflationists like jeff snider, lacy hunt, steven van metre, etc… is that their view is based on how the system is supposed to operate on paper, while your view is based on what is actually happening, regardless of what the law says. I originally dismissed MMT because its most prominent proponents are so wrong about everything else, but after listening to mike green and reading this blog, It seems clear that the MMT view is actually correct. Technically only congress can declare war, but that doesn’t mean iraq or afghanistan didn’t happen.

Leave a reply:

Your email address will not be published.

Site Footer