he's laying the groundwork for the possibility
by in effect its no big deal and he has other stimulus he can use once rates get to zero.....
just so you're not surprised on the next 50bp cut :)
Sorry, but no simple answer here. The structure for whole-laon deals is more typically a full-blown CMO structure (PAC bonds, Z-bonds, companions, floater/inverse pairs, etc.) built out of the AAA cash flows, and then a series of descending rated bonds and finally a first-loss piece. That last piece can run as small a 0.2% on the whole deal [that's not a typo -- 99.8% can be rated at BB or better]. To see that this still works, consider the GNMA program. The US government guarantees all the principal and interest payments for these FHA and VA home buyers that can only afford to put down 5% on their purchase. Before you lose it over "liberal politics", consider that GNMA has turned a regular profit from the tiny 6 basis point fee it takes for this guarantee. GNMA (part of HUD) has just over 100 employees, and outsources everything it can to private contractors. Over the past ten years, it has contributed something over $500 million per year in profits to the US Treasury. If we look at what the market pays for FHA mortgage servicing, that 6 basis point fee has a market value of less than 1/4 of one percent of the principal balance. Frankly, I think this is one of the few things in government that works, since it has brought down the cost of borrowing to buy a house to just 1 1/2 to 2 percent above the cost of US Treasuries, and turned a profit, to boot. That's why every country in the world wants to copy our MBS programs.
Getting back to your A/B question, you can only tell by combing through the company's holdings, and running the bonds through a Bloomberg or similar system. Very time consuming, and requires quite a lot of market experience to judge how to price/predict performance for each piece.
If a company like TMA keeps both pieces, it hasn't removed any risk, but it has extracted the liquidity premium that exists for whole loans if they need to sell. The market is awfully efficient, though, so I would say that the roughly 1/4 point of costs associated with doing the securitization is pretty much equal to the liquidity premium. Hasn't been a free lunch in this area since BankAmerica did their third or fourth deal in the late 70's.
HH: Question about securitized whole loans. If a M REIT, such as TMA were to securitize a package of whole loans they end up, as I understand it, 2 tranches. An A tranche and a B tranche. The credit risk remains in the B tranche first. I think at that point they repo both parts. And then keep both parts in their portfolio.
The question then is how big the B tranche portion is in comparison to the outstnading equity. How much credit risk can they assume before it can wipe out the entire book value? Also, how can you assess the quality of the reserve they put for loan losses?
<<< Here is my question. A LIBOR floater is not a true ARM (there are not enough ARMs to go around), it is a derivative security created from a pool of FRM's that LOOKs like an ARM. >>>
Thanks, hhill1, for the excellent tutorial for us mortgage newbies.
So my question is -- is this type of LIBOR floater what NLY owns, when they say in their 10-Q p. 20 "we held Mortgage-Backed Securities with coupons lined to the one-month and six-month LIBOR ... indices."
I had thought those were MBS composed of actual LIBOR-indexed mortgages, which have been growing in popularity over the past couple of years. Interesting to find out that I might have assumed wrong.
I agree with you! I have learned a lot from this board. Entropy takes a lot of flack but, his input is always informative.I kind of view him as our Devil's advocate! I am so happy that we are back to discussing the pros and cons of Reits rather than the political, racial,etc. postings that we had recently.
The "issuer" is a trust that holds all the MBS backing the deal. Every cent that comes in is paid out to one investor or another. A more descriptive name for these CMOs is multi-class pass-throughs. The rules for distribution are set before the deal closes -- that's what a "structurer" does for a living. Basically, no deal gets done at all unless the value of the derivative bonds is greater than the value of the underlying MBS, plus expenses (the MBS is the collateral of the CMO, or Collateralized Mortgage Obligation).
The buyer of the inverse is typically a hedge fund, but could include a mortgage REIT that wants some hi-octane yield.
There is NO guarantee, even for the PAC bonds, other than 100% of principal over time, and 100% of interest due as long as the bond is outstanding. If prepayments come in so fast that the "companion bonds" -- the floater/inverse pair -- are gone, then the prepayment protection for the PAC bonds is gone.
The thing is, in a pool of thousands of the 6.25% to 6.5% mortgages (the underlying loans in a FNMA 6% MBS), it's statistically impossible for ALL of them to pay off at once. In fact, it's a fair bet that between 5% and 10% will just pay every month for the entire 30 years. Some managers, including the guys at NLY, know this, and will actually pay a pretty good premium for "burned out" premium coupon MBS. For example, you're likely to have lower prepayments today on a FNMA 8% MBS that has already had 90% payoff than you are likely to have on a one-year-old FNMA 6.5% MBS.
Anyway, CMOs exist because 30 years of monthly payments don't fit every investor, while three or five years might, even if you have to accept some principal payment date uncertainty. Given that there's a nice guarantee (for GNMAs, the exact same US Gov't guarantee as US Treasury Notes), you get a pretty good yield pickup for taking this prepay risk. Carving up the risk is what he major MBS bond shops do for their institutional clients.
OK! i just e-mailed it to the generous hhill, but here are my questions:
1. i would like you to confirm or correct my new overview of this subject, based on your excellent post. here it is:
a floating libor is made by taking a pool of FRMs, taking the total income from them, and slicing it up and packaging it in different ways, to suit different buyer's needs. some want a steady payment, guaranteed, for a fixed period of time. they pay extra for this. the FRMs that provide the income may be refi'd, so there is a risk in guaranteeing a steady payment for a fixed period of time. others will buy a 'floating' or changing payment, that changes within certain limits. there must be some intricate planning to convert the fixed incoming stream with this adjustable outgoing stream. the buyer pays extra? the 'issuer' [correct term?]makes a bet on rates rising the right way?
but the overall concept is that the 'issuer' of the floating libor is cutting up this income into different streams. this is what makes it a derivative. if, [purely as an intellectual exercise], all the FRMs were refi'd at once, the 'issuer' would be in big trouble.
2. does the 'issuer' try to come out ahead on this, or break even through hedging or balancing?
3. re: inverse floaters -does the 'issuer' just make a set of pairs, and sell them individually to people who need one or the other? who would need an inverse floater?
<i think my questions are too basic for the board>
Not a chance! There are probably a lot of dummies like me monitoring this board who learn a lot from the interchanges between entropy, bd, and you gentlemen. Thanks for the education and keep it going (online).
<hope this helps,
that was great, thanks. would you mind if i e-mailed you some questions about it? if not, please write me at firstname.lastname@example.org. i think my questions are too basic for the board.
<<Here is my question. A LIBOR floater is not a true ARM (there are not enough ARMs to go around), it is a derivative security created from a pool of FRM's that LOOKs like an ARM>>
You're talking about a floating-rate CMO (legally a REMIC) issued by Fannie or Freddie. These are usually carved up out of regular fixed-rate MBS, so the prepay risk is the re-distributed prepay risk from the underlying mortgage pools. Typically, that prepay risk is more concentrated in the classes that are carved into floater-inverse pairs (they still add up to the same fixed rate), since institutions pay up (i.e., accept lower spread to Treasury) in the scheduled bonds that take less prepay risk. The carry game is played by banks and such by buying Libor floaters that pay a spread like +50, and funding them at Libor flat or +10. They adjust the amount being funded each month as they get the new factor. Works just fine unless spreads widen across the board, or Libor zooms up near the cap on the floater.
Here's a simple example: FNMA 6's are bought at 101 to yield Treasury +180. The CMO structurer runs the pool of MBS at 100 PSA and at 300 PSA to get two vectors of principal payments at a slow and a fast speed. Taking the minimum payment from those two vectors each month, the structurer sets a schedule that allows half of the CMO bonds to fall into a prepay-protected series of PAC (Planned Amortization Class) bonds. Insurers, pension funds, etc. are happy to own these at Treasury +50 at the short end (about 3%), Treasury plus 110 at the long end (about 5.25%). Others, like MBS hedge funds, will take the excess interest from these bonds to yield 10% of 15%. So, that half of the underlying MBS have been sold at 102 or 103. Then the structurer cuts the other half into three floaters at L+50 with a cap of 8% (currently about 1.9% coupon), and one 3-1 inverse floater with a coupon of 6% plus 3*(6-1.9), or about 18.1%. If the PAC and Pac IO bonds were sold at a high enough price to lower the effective cost of the floater/inverse pair below par, the dealer makes a profit by selling both of those classes at par, so the matched-funders make a profit on the Libor floaters as long as the Libor rate stays below 7.5%, and the high-risk buyer makes a double-digit yield as long as Libor doesn't go too much above 4%.
Both the floater and the inverse buyers are accepting about double the prepay volatility as the PAC and Pac IO buyers take.
hope this helps,