With the question of the Fed's portfolio normalization now all the rage, accentuated by yesterday's FOMC Minutes announcement that runoff could start later this year - even as many traders admit nobody has any idea what will happen if and when the Fed starts reducing its holdings, mostly of MBS - on Thursday the NY Fed, the Fed's trading desk, provided a glimpse into its thinking on how this will play out in its latest Domestic Market Operations annual report.
According to the report, the Fed's bond holdings could drop to about $2.8 trillion by the end of 2021 - a $1.7 trillion reduction over the next 5 years - with the New York Fed now projecting its balance sheet will reach a "normalized" state some two quarter earlier however with approximately $600 billion more assets than in a year-ago estimate. The U.S. central bank currently has some $4.5 trillion in Treasury and mortgage bonds.
To be sure, many things can and will happen between now and 2021, including the US may have a new president.
Which is why what we found more interesting was the NY Fed's own forecast on the start of renormalization, which disagreed with the FOMC Minutes, in that Bill Dudley's Fed does not expect the Fed to start "renormalizing" until mid-2018, to wit: "the size of the SOMA portfolio is projected to remain largely unchanged at its current level of approximately $4.2 trillion through mid-2018, while full reinvestments continue."
What happens to the balance sheet then:
After that date, it starts to decline as reinvestments are phased out and then ended altogether in mid-2019. The Federal Reserve’s securities holdings then decline until the portfolio reaches its normalized size in the fourth quarter of 2021 (Chart 26). At that time, the domestic securities portfolio is estimated to be about $2.8 trillion, with a slightly higher concentration in Treasury securities than in agency MBS. Thereafter, Treasury-driven growth of securities holdings supports trend balance sheet growth, and agency debt and agency MBS holdings continue to run off.
The NY Fed on suspension of reinvestments vs outright selling:
Once the FOMC ends reinvestments, the pace of the reduction in the size of the SOMA portfolio will largely be driven by the pace of principal receipts from SOMA securities holdings (Chart 27). The timing of principal payments from maturing Treasury securities and agency debt securities is a known function of current SOMA holdings. In contrast, projected principal pay-downs associated with agency MBS are model-based estimates that are subject to considerable uncertainty because of the embedded prepayment option. The actual pay-down path will depend on a variety of factors, including the path of interest rates, changes in housing prices, credit conditions, and other government policy initiatives.
Finally, how the latest forecast differs from last years:
The point of normalization in late 2021 is projected to occur almost two quarters earlier than in the 2015 baseline (Chart 28). The balance sheet starts to contract just over a year later than it was expected to in the 2015 baseline given a longer-than-previously anticipated period for reinvestments to continue. (The December 2015 baseline was modeled on an assumption that reinvestments would begin to be phased out in the first half of 2017.) However, a larger long-run balance sheet size in the current baseline, driven by the assumption about a higher level of reserve balance liabilities in a future policy implementation framework, requires less of the portfolio to run off once such a contraction starts.
And some parting words:
Of course, banks’ demand for reserves and the level of reserves the FOMC will choose to maintain in its long-run policy implementation framework remain uncertain. A set of alternative scenarios highlights the sensitivity of SOMA portfolio balances to different long-run levels of Federal Reserve liabilities. These scenarios illustrate the degree to which increases (decreases) in liabilities imply a larger (smaller) level of the SOMA in the long run and how long it might take to achieve a normalized portfolio size. While the projections are modeled with regard to alternative levels of reserve balances, the specific type of liability is not material; the effect on SOMA portfolio balances would be similar if the alternative levels of liabilities arose from changes in other line items, such as Federal Reserve notes, the TGA, the foreign repo pool, or DFMU balances.
Under a scenario in which reserve balances are $100 billion in the long run (the baseline in prior years’ reports), the size of the balance sheet is normalized in the fourth quarter of 2022, approximately one year later than in the baseline scenario (Chart 29). In contrast, under a scenario in which reserves are $1 trillion in the long run, the size of the balance sheet is normalized in the fourth quarter of 2020, nearly one year sooner than in the baseline. Given that Treasury purchases resume at an earlier date, by the end of the forecast horizon the portfolio is more heavily weighted to Treasury securities than it is in the baseline scenario.
In other words, if all goes according to plan, the Fed will consider its "renormalization" mission complete in about 5 years, at which point it will have no qualms about launching even more QE if it has to.