30-yr TBAs

If there existed a universal rule of thumb in upgrading empirical prepay models, it would state that the option cost should go up after an upgrade. This is because empirical models reflect historical trends that normally witness an improved refinancibility. At the dawn of MBS, refinancing was a hassle; today it may lie at an applicant’s mouse click (the reverse trend is not impossible and may come with newly invented fees and taxes).

Our “drift of models” is not an exception and confirms the rule (Exhibit 1).

Exhibit 1

TBA	Price	OAS			Z-spread			Option Cost
		old	new	diff	old	new	diff	old	new	diff
FNCL4.5	95.64	4.5	1.9	-2.6	42.4	48.1	5.7	38.0	46.2	8.2
FNCL5.0	98.17	1.2	-0.1	-1.3	56.9	61.6	4.7	55.7	61.7	6.0
FNCL5.5	100.23	3.0	0.9	-2.1	75.0	75.9	0.9	72.0	75.0	3.0
FNCL6.0	101.86	5.7	-3.0	-8.7	78.3	79.5	1.2	72.6	82.5	9.9
FNCL6.5	103.11	33.6	16.0	-17.6	88.2	74.8	-13.4	54.6	58.8	4.2
FNCL7.0	104.72	44.4	30.5	-13.9	79.2	69.8	-9.5	34.8	39.2	4.5
FNCL7.5	105.81	44.3	29.5	-14.9	46.8	42.4	-4.4	2.4	12.9	10.5
FNCL8.0	106.88	47.4	45.4	-2.1	39.6	49.5	9.9	-7.8	4.1	11.9

As seen in Exhibit 1, the option cost is up and the OAS is down for all coupons once we transitioned from “old” OAS model 5.2d (prepay model 4.3.4) to the “new” OAS model 6.0 (prepay model 5.1). The Z-spreads got lower for most premiums, but higher for discounts. The former fact suggests that the premiums do refinance faster in the new model (see confirmation on Exhibit 2); all the coupons are more rate-sensitive, which implies a higher prepay option cost.

>>>