Prepayments were relatively flat for the month of June, slowing after the increases posted in May. This was expected as mortgage rates climbed to 52 weeks highs during the month. We saw slightly increased speeds from FNMA 30 discounts, about 4-6%, and slightly decreased speeds for premiums, at about -5%. FHLMC 30-year collateral posted decreases across the board at about the same -5% magnitude seen in the FNMA premiums. This pattern was consistent for both FHLMC & FNMA 15-year collateral as well. Prepayments for the GNMAs showed the same behavior, with speeds rising slightly for discounts and falling for premiums. We show actual pool CPR speeds of selected coupon buckets along with corresponding model forecasts in the tables below.
v4.3.4a
FNMA 15 |
FNMA 30 |
FHLMC 15 |
|||||||
Coupon |
Actual |
Model |
Current Balance |
Actual |
Model |
Current Balance |
Actual |
Model |
Current Balance |
4.0 |
7.8 |
11.47 |
$26,912,331,958 |
6.66 |
4.95 |
$557,504,640 |
8.44 |
11.25 |
$40,639,134,184 |
4.5 |
9.42 |
11.66 |
$106,805,054,490 |
8.08 |
4.52 |
$40,710,395,309 |
9.26 |
11.11 |
$130,284,062,345 |
5.0 |
11.09 |
12.54 |
$111,230,578,840 |
9.52 |
4.26 |
$338,479,277,896 |
10.54 |
10.96 |
$126,127,204,721 |
5.5 |
12.59 |
12.91 |
$43,885,779,997 |
12.04 |
5.92 |
$443,969,507,424 |
11.63 |
10.93 |
$51,519,624,150 |
6.0 |
15.88 |
16.8 |
$21,168,552,608 |
14.88 |
11.46 |
$197,969,210,381 |
15.63 |
16.38 |
$21,721,891,062 |
6.5 |
18 |
20.48 |
$7,338,120,636 |
19.83 |
15.47 |
$70,359,394,426 |
17.87 |
19.4 |
$7,770,309,524 |
7.0 |
18.95 |
23.22 |
$2,448,171,279 |
24.47 |
21.52 |
$23,273,661,620 |
19.23 |
22.34 |
$2,827,807,427 |
7.5 |
19.98 |
24.86 |
$931,716,173 |
24.21 |
25.54 |
$7,640,977,340 |
20.7 |
23.99 |
$1,167,603,625 |
8.0 |
22.7 |
27.07 |
$328,573,007 |
23.21 |
30.3 |
$3,486,877,875 |
21.75 |
25.78 |
$377,505,348 |
FHLMC 30 |
GNMA 15 |
GNMA 30 |
|||||||
Coupon |
Actual
|
Model
|
Current Balance
|
Actual
|
Model
|
Current Balance |
Actual
|
Model
|
Current Balance
|
4.0 |
3.81 |
4.94 |
$545,616,241 |
13.18 |
8.35 |
$1,098,448,550 |
20.05 |
6.57 |
$2,485,649,337 |
4.5 |
6.88 |
3.62 |
$44,738,520,111 |
12.4 |
8.5 |
$3,953,247,286 |
16.35 |
7.17 |
$9,495,025,799 |
5.0 |
8.53 |
3.62 |
$365,665,371,435 |
14.08 |
9.32 |
$6,538,675,325 |
14.13 |
8.91 |
$67,798,664,767 |
5.5 |
10.82 |
4.81 |
$393,512,357,545 |
14.15 |
10.34 |
$2,786,280,150 |
14.98 |
14.04 |
$111,912,988,084 |
6.0 |
13.02 |
8.7 |
$166,537,056,693 |
15.79 |
13.55 |
$1,796,404,108 |
16.56 |
17.73 |
$60,583,476,849 |
6.5 |
20.31 |
14.82 |
$58,939,204,225 |
18.6 |
14.51 |
$836,516,902 |
20.51 |
22.16 |
$23,870,772,171 |
7.0 |
20.64 |
17.38 |
$18,044,580,427 |
19.85 |
15.56 |
$522,533,516 |
21.52 |
22.33 |
$12,117,960,552 |
7.5 |
21.17 |
22.7 |
$6,904,010,915 |
20.54 |
18.3 |
$201,235,202 |
21.79 |
26.31 |
$5,105,594,228 |
8.0 |
19.71 |
27.89 |
$3,287,192,185 |
18.59 |
21.32 |
$103,722,612 |
20.33 |
33.41 |
$3,094,014,020 |
v5.1e
FNMA 15 |
FNMA 30 |
FHLMC 15 |
|||||||
Coupon |
Actual
|
Model
|
Current Balance
|
Actual
|
Model
|
Current Balance |
Actual
|
Model
|
Current Balance
|
4.0 |
7.8 |
7.39 |
$26,912,331,958 |
6.66 |
6.97 |
$557,504,640 |
8.44 |
7.45 |
$40,639,134,184 |
4.5 |
9.42 |
8.03 |
$106,805,054,490 |
8.08 |
7.05 |
$40,710,395,309 |
9.26 |
7.59 |
$130,284,062,345 |
5.0 |
11.09 |
9.72 |
$111,230,578,840 |
9.52 |
8.63 |
$338,479,277,896 |
10.54 |
8.62 |
$126,127,204,721 |
5.5 |
12.59 |
11.34 |
$43,885,779,997 |
12.04 |
10.76 |
$443,969,507,424 |
11.63 |
9.79 |
$51,519,624,150 |
6.0 |
15.88 |
13.74 |
$21,168,552,608 |
14.88 |
12.85 |
$197,969,210,381 |
15.63 |
13.5 |
$21,721,891,062 |
6.5 |
18 |
16.17 |
$7,338,120,636 |
19.83 |
16.05 |
$70,359,394,426 |
17.87 |
15.54 |
$7,770,309,524 |
7.0 |
18.95 |
18.78 |
$2,448,171,279 |
24.47 |
21.23 |
$23,273,661,620 |
19.23 |
17.22 |
$2,827,807,427 |
7.5 |
19.98 |
20.75 |
$931,716,173 |
24.21 |
26.01 |
$7,640,977,340 |
20.7 |
19.43 |
$1,167,603,625 |
8.0 |
22.7 |
22.58 |
$328,573,007 |
23.21 |
30.03 |
$3,486,877,875 |
21.75 |
22.02 |
$377,505,348 |
FHLMC 30 |
GNMA 15 |
GNMA 30 |
|||||||
Coupon |
Actual
|
Model
|
Current Balance
|
Actual
|
Model
|
Current Balance |
Actual
|
Model
|
Current Balance
|
4.0 |
3.81 |
7.12 |
$545,616,241 |
13.18 |
11.93 |
$1,098,448,550 |
20.05 |
8.43 |
$2,485,649,337 |
4.5 |
6.88 |
6.76 |
$44,738,520,111 |
12.4 |
11.93 |
$3,953,247,286 |
16.35 |
11.2 |
$9,495,025,799 |
5.0 |
8.53 |
8.14 |
$365,665,371,435 |
14.08 |
13.26 |
$6,538,675,325 |
14.13 |
14.17 |
$67,798,664,767 |
5.5 |
10.82 |
9.89 |
$393,512,357,545 |
14.15 |
13.59 |
$2,786,280,150 |
14.98 |
16.21 |
$111,912,988,084 |
6.0 |
13.02 |
11.12 |
$166,537,056,693 |
15.79 |
15.17 |
$1,796,404,108 |
16.56 |
18.09 |
$60,583,476,849 |
6.5 |
20.31 |
15.66 |
$58,939,204,225 |
18.6 |
16.83 |
$836,516,902 |
20.51 |
22.73 |
$23,870,772,171 |
7.0 |
20.64 |
19.53 |
$18,044,580,427 |
19.85 |
17.12 |
$522,533,516 |
21.52 |
23.58 |
$12,117,960,552 |
7.5 |
21.17 |
24 |
$6,904,010,915 |
20.54 |
17.01 |
$201,235,202 |
21.79 |
24.52 |
$5,105,594,228 |
8.0 |
19.71 |
28.38 |
$3,287,192,185 |
18.59 |
17.29 |
$103,722,612 |
20.33 |
25.05 |
$3,094,014,020 |
Recent market dynamics have led us to closely examine the SATO (Spread-at- Origination) effect. SATO measures the spread between the coupon on the loan and the prevailing market rate at the time of origination and was once an important effect in the agency models. This effect, however, seems to be fading over time. The initial period in which prepayments are dampened because of a perceived borrower impairment to refinance is not only getting shorter, but the actual level of prepayments for borrowers holding higher SATO loans has been rising over time. As Alex Levin discusses in this month's Valuation Commentary, the model effect for SATO was accurate in 2002, but, now, a tuning of 0.33 for newly originated SATO loans greatly improves model accuracy in 2006.
Why is this change occurring? In agency pools, credit quality is usually very good, and the reason a borrower is receiving a higher SATO loan is a little less unclear now as compared to just a few years ago. With the boom in sub-prime and alt-a lending over the last couple years, borrowers with impaired credit have been able to secure loans at much better rates and at a lower cost for their credit class than in the past.
We still observe noticeably dampened prepayments for a period of time at the beginning of a loan's life in these non-agency pools; so we know, fundamentally, that the SATO effect is still a factor. It just does not appear to be as strong as it once was in agency pools. We will look more closely at this factor in our next model re-estimation, making updates to reflect the changing dynamic of the agency market.
