In your first example of GNMA II, why would you add the balance reduction of 4.4bn to the net purchase of 4.84bn? The latter purchase increases the balance, no?
Its more adding the net purchase settlement of 4.84b to the balance reduction of 4.4b. Yes. the 4.84 purchase settlement increases the balance, so if it didnt happen that week the balance would have been 4.84b lower than was reported, yielding straight up the true payment of 9.24b just based on balance differential from week before to week of the pyament.
the Fed projects the runoff for MBS's to start at $22.5B/mo but by end of 2024, it will be only $18.5B/mo. I would have expected it to increase because of the way mortgages amortize. Could you explain why it decreases?
As a mortgage gets later in its amortization the principal portion of it does increase. That said it will be a pretty minimal effect in the early years of a 30 year (reflective of most of FED MBS). Also the regular monthly payment aspect of MBS payments is ~5b a month, so its relatively small impact on the total MBS payment vs. prepayment activity. The FEDs projections seem to more anticipate prepayment activity continuing to slow over the next 2 years. I generally agree with respect to refinances which i think will remain super low next 2 years. Prepayments due to purchases or defaults imo will depend on how quickly prices reset and how much supply comes on line due to distressed selling by investors. We shall wee
Hi John, great post. I want to ask how would FED close the gap between projected runoff in MBS i.e. 35B and expected principal payments i.e. 23B (as per your latest post)? Is FED going to reduce some other security off its balance sheet as it does in the case of UST coupons by reducing UST Bills?
No. As a result only the 23b (or whatever the principal payments received in September end up being) will be rolled off the B/S. At some point the FED could decide to start actively selling MBS to bridge some or all of that cap, but they have not announced any plan to do so of yet and are unlikely to imo.
In your first example of GNMA II, why would you add the balance reduction of 4.4bn to the net purchase of 4.84bn? The latter purchase increases the balance, no?
Its more adding the net purchase settlement of 4.84b to the balance reduction of 4.4b. Yes. the 4.84 purchase settlement increases the balance, so if it didnt happen that week the balance would have been 4.84b lower than was reported, yielding straight up the true payment of 9.24b just based on balance differential from week before to week of the pyament.
According to Figure 3 in:
https://www.richmondfed.org/publications/research/economic_brief/2022/eb_22-15
the Fed projects the runoff for MBS's to start at $22.5B/mo but by end of 2024, it will be only $18.5B/mo. I would have expected it to increase because of the way mortgages amortize. Could you explain why it decreases?
As a mortgage gets later in its amortization the principal portion of it does increase. That said it will be a pretty minimal effect in the early years of a 30 year (reflective of most of FED MBS). Also the regular monthly payment aspect of MBS payments is ~5b a month, so its relatively small impact on the total MBS payment vs. prepayment activity. The FEDs projections seem to more anticipate prepayment activity continuing to slow over the next 2 years. I generally agree with respect to refinances which i think will remain super low next 2 years. Prepayments due to purchases or defaults imo will depend on how quickly prices reset and how much supply comes on line due to distressed selling by investors. We shall wee
Hi John, great post. I want to ask how would FED close the gap between projected runoff in MBS i.e. 35B and expected principal payments i.e. 23B (as per your latest post)? Is FED going to reduce some other security off its balance sheet as it does in the case of UST coupons by reducing UST Bills?
No. As a result only the 23b (or whatever the principal payments received in September end up being) will be rolled off the B/S. At some point the FED could decide to start actively selling MBS to bridge some or all of that cap, but they have not announced any plan to do so of yet and are unlikely to imo.
Would you consider this approach to be valid for calculating the MBS purchased between May 13th and June 13th?
https://gist.github.com/dharmatech/b7b2757bc8d1e42b06e08704b22c78bd
It lines up with the $34.6 on announced by the Fed.
Great post John!
I have a question that was too large to fit in these comments:
https://gist.github.com/dharmatech/516f3e6919b98c4c3d6ee6933d1970f8
It's regarding the use of the spreadsheet data available here:
https://www.newyorkfed.org/markets/ambs/ambs_schedule
Basically I'm wondering, did you just download all of them into a database to query as needed?
Nevermind... I see that you mention doing exactly this in the technical appendix. 🙂